Re: [HACKERS] upper planner path-ification
Tom == Tom Lane t...@sss.pgh.pa.us writes: Tom Hm. That's a hangover from when query_planner also gave back a Tom Plan (singular) rather than a set of Paths. I don't see any Tom fundamental reason why we couldn't generalize it to be a list of Tom potentially useful output orderings rather than just one. But I'm Tom a bit concerned about the ensuing growth in planning time; is it Tom really all that useful? The planning time growth is a possible concern, yes. The potential gain is eliminating one sort step, in the case when the input has a usable sorted path but grouping_planner happens not to ask for it (when there's more than just a single rollup, the code currently asks for one of the sort orders pretty much arbitrarily since it has no real way to know otherwise). Whether that would justify it... I don't know. Maybe that's one to save for later to see if there's any feedback from actual use. -- Andrew (irc:RhodiumToad) -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] jsonb concatenate operator's semantics seem questionable
On 20/05/15 01:38, Jim Nasby wrote: On 5/18/15 3:15 PM, Marko Tiikkaja wrote: On 2015-05-18 22:10, Josh Berkus wrote: On 05/18/2015 01:04 PM, Ryan Pedela wrote: In the context of splitting shallow and deep merge into two operators, I think + is better for shallow and || better for deep. The reason for + is because many programming languages have this behavior. If I see the below code in language I have never used before: objC = objA + objB My default assumption is that + performs a shallow merge. Like I said, I would rather there just be one operator. Thank you, that helps. Anyone else? If everyone thinks the operators mean different things, we could just not add any operators and only provide functions instead. My $0.02: I would expect || to be what I want to use to add something to an existing JSON document, no matter what the path of what I'm adding is. In other words, deep merge. I certainly wouldn't expect it to be shallow. If we get this wrong now, we'll be stuck with it forever. At a minimum I think we should use anything other than || until we can figure this out. That leaves || available for whichever case we decide on. I am of strong opinion that concat should be shallow by default. Again it's how jquery works by default, it's how python's dict.update works and you can find this behavior in other languages as well when dealing with nested hashes. It's also how json would behave if you'd just did string concatenation (removing the outermost curly brackets) and parse it to json afterwards. I think this whole discussion shows primarily that it's by far not universally agreed if concatenation of json should be shallow or deep by default and AFAICS this is true even in javascript world so we don't really have where to look for precedents. Given the above I would vote to just provide the function and leave out the || operator for now. -- Petr Jelinek http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training Services -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Change pg_cancel_*() to ignore current backend
On 5/19/15 9:19 PM, Fabrízio de Royes Mello wrote:
We could add a second parameter to the current functions:
allow_own_pid DEFAULT false. To me that seems better than an
entirely separate set of functions.
+1 to add a second parameter to current functions.
Instead of allow_own_pid, I went with skip_own_pid. I have the function
still returning true even when it skips it's own PID... that seems a bit
weird, but I think it's better than returning false. Unless someone
thinks it should return NULL, but I don't see that as any better either.
--
Jim Nasby, Data Architect, Blue Treble Consulting
Data in Trouble? Get it in Treble! http://BlueTreble.com
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 89a609f..b405876 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -16508,7 +16508,7 @@ SELECT set_config('log_statement_stats', 'off', false);
tbody
row
entry
-literalfunctionpg_cancel_backend(parameterpid/parameter
typeint/)/function/literal
+literalfunctionpg_cancel_backend(parameterpid/parameter
typeint/optional, parameterskip_my_pid/ typeboolean/
/optional)/function/literal
/entry
entrytypeboolean/type/entry
entryCancel a backend's current query. This is also allowed if the
@@ -16532,7 +16532,7 @@ SELECT set_config('log_statement_stats', 'off', false);
/row
row
entry
-literalfunctionpg_terminate_backend(parameterpid/parameter
typeint/)/function/literal
+literalfunctionpg_terminate_backend(parameterpid/parameter
typeint/optional, parameterskip_my_pid/ typeboolean/
/optional)/function/literal
/entry
entrytypeboolean/type/entry
entryTerminate a backend. This is also allowed if the calling role
@@ -16562,6 +16562,10 @@ SELECT set_config('log_statement_stats', 'off', false);
The role of an active backend can be found from the
structfieldusename/structfield column of the
structnamepg_stat_activity/structname view.
+
+ There is an optional second parameter of type typeboolean/type. If
+ literaltrue/ (the default), functionpg_cancel_backend/ and
+ functionpg_terminate_backend/ will not signal the current backend.
/para
para
diff --git a/src/backend/catalog/system_views.sql
b/src/backend/catalog/system_views.sql
index 18921c4..a0cc975 100644
--- a/src/backend/catalog/system_views.sql
+++ b/src/backend/catalog/system_views.sql
@@ -869,6 +869,14 @@ COMMENT ON FUNCTION ts_debug(text) IS
--
CREATE OR REPLACE FUNCTION
+ pg_cancel_backend(pid int, skip_my_pid boolean DEFAULT true)
+ RETURNS boolean STRICT VOLATILE LANGUAGE internal AS 'pg_cancel_backend';
+
+CREATE OR REPLACE FUNCTION
+ pg_terminate_backend(pid int, skip_my_pid boolean DEFAULT true)
+ RETURNS boolean STRICT VOLATILE LANGUAGE internal AS 'pg_terminate_backend';
+
+CREATE OR REPLACE FUNCTION
pg_start_backup(label text, fast boolean DEFAULT false)
RETURNS pg_lsn STRICT VOLATILE LANGUAGE internal AS 'pg_start_backup';
diff --git a/src/backend/utils/adt/misc.c b/src/backend/utils/adt/misc.c
index 61d609f..dce8498 100644
--- a/src/backend/utils/adt/misc.c
+++ b/src/backend/utils/adt/misc.c
@@ -94,8 +94,12 @@ current_query(PG_FUNCTION_ARGS)
#define SIGNAL_BACKEND_NOPERMISSION 2
#define SIGNAL_BACKEND_NOSUPERUSER 3
static int
-pg_signal_backend(int pid, int sig)
+pg_signal_backend(int pid, int sig, bool skip_own_pid)
{
+ /* Skip our own pid unless we're told not to */
+ if (skip_own_pid pid == MyProcPid)
+ return SIGNAL_BACKEND_SUCCESS;
+
PGPROC *proc = BackendPidGetProc(pid);
/*
@@ -158,7 +162,7 @@ pg_signal_backend(int pid, int sig)
Datum
pg_cancel_backend(PG_FUNCTION_ARGS)
{
- int r = pg_signal_backend(PG_GETARG_INT32(0),
SIGINT);
+ int r = pg_signal_backend(PG_GETARG_INT32(0),
SIGINT, PG_GETARG_BOOL(1));
if (r == SIGNAL_BACKEND_NOSUPERUSER)
ereport(ERROR,
@@ -182,7 +186,7 @@ pg_cancel_backend(PG_FUNCTION_ARGS)
Datum
pg_terminate_backend(PG_FUNCTION_ARGS)
{
- int r = pg_signal_backend(PG_GETARG_INT32(0),
SIGTERM);
+ int r = pg_signal_backend(PG_GETARG_INT32(0),
SIGTERM, PG_GETARG_BOOL(1));
if (r == SIGNAL_BACKEND_NOSUPERUSER)
ereport(ERROR,
diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h
index b5b9345..475545b 100644
--- a/src/include/catalog/pg_proc.h
+++ b/src/include/catalog/pg_proc.h
@@ -3128,9 +3128,9 @@ DESCR(get OID of current session's temp schema, if any);
DATA(insert OID = 2855 ( pg_is_other_temp_schema PGNSP PGUID 12 1 0 0 0
f f f f t f s 1 0 16 26 _null_ _null_ _null_ _null_ _null_
pg_is_other_temp_schema _null_ _null_ _null_ ));
DESCR(is schema another session's temp schema?);
-DATA(insert OID = 2171 ( pg_cancel_backend PGNSP PGUID 12 1 0 0
Re: [HACKERS] Minor ON CONFLICT related fixes
On Tue, May 19, 2015 at 2:23 PM, Andres Freund and...@anarazel.de wrote:
Pushed.
I eyeballed the commit, and realized that I made a trivial error. New
patch attached fixing that.
Sorry for not getting this fix completely right first time around.
Don't know how I missed it.
--
Peter Geoghegan
diff --git a/src/backend/utils/adt/ruleutils.c b/src/backend/utils/adt/ruleutils.c
index 8cdef08..0585251 100644
--- a/src/backend/utils/adt/ruleutils.c
+++ b/src/backend/utils/adt/ruleutils.c
@@ -5500,7 +5500,7 @@ get_insert_query_def(Query *query, deparse_context *context)
get_rule_expr(confl-arbiterWhere, context, false);
}
}
- else
+ else if (confl-constraint != InvalidOid)
{
char *constraint = get_constraint_name(confl-constraint);
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] CTE optimization fence on the todo list?
I need this feature a lot. Can anyone point me to a place in the code where I can hack together a quick-and-dirty, compatibility-breaking implementation? Thanks! On Sun, May 3, 2015 at 10:03 PM, Jim Nasby jim.na...@bluetreble.com wrote: On 5/3/15 11:59 AM, Andrew Dunstan wrote: On 05/03/2015 11:49 AM, Tom Lane wrote: Andrew Dunstan and...@dunslane.net writes: On 05/01/2015 07:24 PM, Josh Berkus wrote: (A possible compromise position would be to offer a new GUC to enable/disable the optimization globally; that would add only a reasonably small amount of control code, and people who were afraid of the change breaking their apps would probably want a global disable anyway.) This could be a very bad, almost impossible to catch, behaviour break. Even if we add the GUC, we're probably going to be imposing very significant code audit costs on some users. On what grounds do you claim it'd be a behavior break? It's possible that the subquery flattening would result in less-desirable plans not more-desirable ones, but the results should still be correct. I meant w.r.t. performance. Sorry if that wasn't clear. To put this in perspective... I've seen things like this take query runtime from minutes to multiple hours or worse; bad enough that behavior break becomes a valid description. We definitely need to highlight this in the release notes, and I think the GUC would be mandatory. -- Jim Nasby, Data Architect, Blue Treble Consulting Data in Trouble? Get it in Treble! http://BlueTreble.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Making the regression tests halt to attach a debugger
On 5/18/15 12:15 PM, Andrew Dunstan wrote: On 05/18/2015 01:05 PM, Tom Lane wrote: Meh. You could also add select pg_backend_pid() or some such. But really, the way I generally do this is to run gdb via a script that auto-attaches to the right postgres process if at all possible. Removes the whole problem. This should go on the wiki. https://wiki.postgresql.org/wiki/Gdblive_script; and linked from gdb section of Developer FAQ. -- Jim Nasby, Data Architect, Blue Treble Consulting Data in Trouble? Get it in Treble! http://BlueTreble.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On Tue, May 19, 2015 at 2:28 PM, Simon Riggs si...@2ndquadrant.com wrote: On 19 May 2015 at 17:10, Peter Geoghegan p...@heroku.com wrote: On Tue, May 19, 2015 at 1:57 PM, Simon Riggs si...@2ndquadrant.com wrote: We should allow DO UPDATE to exclude a constraint and apply a deterministic order to the constraints. 1. PK if it exists. 2. Replica Identity, when not PK, 3. UNIQUE constraints in name order, like triggers, so users can define a default evaluation order, just like they do with triggers. That seems like something way worse than just allowing it for all constraints. I'm talking about the evaluation order; it would still match all constraints, otherwise they wouldn't be constraints. But it doesn't match all constraints when a would-be conflict is detected. IOW, we lock the row and go to UPDATE, and then the user is on their own insofar as avoiding duplicate violations goes. What might have happened in other unique indexes (had that original would-be dup violation not occurred) is irrelevant (with the MySQL thing, say) -- you better just get it right, and know that if a dup violation occurs it was the one you anticipated (e.g. because there is only one unique index anyway). With Postgres, we want to make sure that the user has put thought into the condition they take that update path on, and so it is mandatory (it can infer multiple unique indexes, but only when they're basically equivalent for this purpose). I think I agree with you, though: We should change things so that the relcache gives indexes in something like the ordering that you outline, rather than in the current arbitrary (though consistent) OID order. However, I think that this should be done to avoid unnecessary index bloat (fail early), and I don't think it makes much sense to do it on the grounds you outline. This is because you can still easily take the alternative path for the wrong reason, causing subtle logical corruption. You can still not match all indexes because one index had a would-be dup violation (and so, as I said, it doesn't matter what would have happened with the other ones). Maybe you still get a dup violation from the update, saving you, but who wants to rely on that? 2) Compatibility with MySQL But what you describe isn't compatible with MySQL. It's totally novel. Upthread you said It's trivial to modify Postgres to not require that a specific unique index be inferred, so that you can omit the inference specification for DO UPDATE just as you can for DO NOTHING. That would make it work in a similar way to MySQL Similar is good and useful. Full compatibility is even better. I actually do not feel strongly that it would be terrible to allow the user to omit an inference clause for the DO UPDATE variant (on the grounds of that being closer to MySQL). After all, we don't mandate that the user specifies an explicit targetlist for INSERT, and that seems like a footgun to me. If you want to make the case for doing things that way, I probably will not oppose it. FWIW, I don't think it's unreasonable to have a little discussion on fine points of semantics like that post feature-freeze. -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Change pg_cancel_*() to ignore current backend
On 2015-05-20 00:59, Jim Nasby wrote: I find it annoying to have to specifically exclude pg_backend_pid() from pg_stat_activity if I'm trying to kill a bunch of backends at once, and I can't think of any reason why you'd ever want to call a pg_cancel_* function with your own PID. That's a rather easy way of testing that you're handling FATAL errors correctly from a driver/whatever. .m -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Change pg_cancel_*() to ignore current backend
On 5/19/15 6:30 PM, David G. Johnston wrote: On Tue, May 19, 2015 at 4:23 PM, Marko Tiikkaja ma...@joh.to mailto:ma...@joh.towrote: On 2015-05-20 00:59, Jim Nasby wrote: I find it annoying to have to specifically exclude pg_backend_pid() from pg_stat_activity if I'm trying to kill a bunch of backends at once, and I can't think of any reason why you'd ever want to call a pg_cancel_* function with your own PID. That's a rather easy way of testing that you're handling FATAL errors correctly from a driver/whatever. I'm having trouble thinking of a PC name for the function we create that should do this; while changing the pg_cancel_* functions to operate more safely. We could add a second parameter to the current functions: allow_own_pid DEFAULT false. To me that seems better than an entirely separate set of functions. -- Jim Nasby, Data Architect, Blue Treble Consulting Data in Trouble? Get it in Treble! http://BlueTreble.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On Tue, May 19, 2015 at 5:03 PM, Tom Lane t...@sss.pgh.pa.us wrote: Peter Geoghegan p...@heroku.com writes: I think I agree with you, though: We should change things so that the relcache gives indexes in something like the ordering that you outline, rather than in the current arbitrary (though consistent) OID order. I'm fairly sure that there are aspects of the code that rely on indexes being returned by RelationGetIndexList() in a stable order. While I doubt that has to be exactly increasing-OID-order, I'm quite concerned about allowing the order to depend on mutable aspects of the indexes, like names. I thought the importance of the ordering was just down to some AMs (like hash) using heavyweight locks. This could cause unprincipled deadlocks in the face of an inconsistent ordering. nbtree used to use page-level heavyweight locks many years ago, too, so this used to be a big, obvious requirement. Maybe there is another reason, but AFAICR there are no hints of that from the relevant code, and I've looked carefully. If it was ever changed, I think it could be done in a way that didn't add any problems, assuming I've accounted for all the ways in which changing the ordering could be problematic. -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
* Simon Riggs (si...@2ndquadrant.com) wrote: On 19 May 2015 at 16:49, Robert Haas robertmh...@gmail.com wrote: On Tue, May 19, 2015 at 3:00 PM, Simon Riggs si...@2ndquadrant.com wrote: As long as the cookie is randomly generated for each use, then I don't see a practical problem with that approach. If the client sets the cookie via an SQL command, that command would be written to the log, and displayed in pg_stat_activity. A malicious user might be able to get it from one of those places. A malicious user might also be able to just guess it. I don't really want to create a situation where any weakess in pgpool's random number generation becomes a privilege-escalation attack. A protocol extension avoids all of that trouble, and can be target for 9.6 just like any other approach we might come up with. I actually suspect the protocol extension will be FAR easier to fully secure, and thus less work, not more. That's a reasonable argument. So +1 to protocol from me. To satisfy Tom, I think this would need to have two modes: one where the session can never be reset, for ultra security, and one where the session can be reset, which allows security and speed of pooling. For my 2c, I continue to agree with a protocol-based approach, but I don't think having two modes would actually satisfy concerns regarding the security- we're still going to have to fix any issues which are security related that come up from having the session able to be reset mode. That said, we know connection poolers are already using SET SESSION AUTH (which is clearly far worse than what we're proposing to do here..) and clearly we support SET ROLE, so any issues with those methods really should be getting addressed anyway. Perhaps we can continue to beg off in the SET SESSION AUTH case by hiding behind you're a superuser or you're using it wrong but that doesn't actually make anyone more secure and we clearly need to address the SET ROLE case, as that is absolutely expected to work correctly. As for the discussion regarding having a connection pooler built-in- that is absolutely something we need to do, in my view, because any external connection pooler isn't going to offer the same set of capabilities that core does and we continue to fight with the concerns around changing the wireline protocol which hamstrings our progress in this area. That isn't to say it's all roses if we just built it in, because clearly it's not and there's work to be done there, but a connection pooler which is tied closely to core and which is upgraded and deployed with it could be much more easily changed and improved. On the other hand, I'd really like to see improvement to our protocol too and perhaps this is a way to get those, though it hasn't been happening so far, unfortunately. Thanks! Stephen signature.asc Description: Digital signature
Re: [HACKERS] Change pg_cancel_*() to ignore current backend
On Tue, May 19, 2015 at 4:23 PM, Marko Tiikkaja ma...@joh.to wrote: On 2015-05-20 00:59, Jim Nasby wrote: I find it annoying to have to specifically exclude pg_backend_pid() from pg_stat_activity if I'm trying to kill a bunch of backends at once, and I can't think of any reason why you'd ever want to call a pg_cancel_* function with your own PID. That's a rather easy way of testing that you're handling FATAL errors correctly from a driver/whatever. I'm having trouble thinking of a PC name for the function we create that should do this; while changing the pg_cancel_* functions to operate more safely. David J.
Re: [HACKERS] jsonb concatenate operator's semantics seem questionable
On 5/18/15 3:15 PM, Marko Tiikkaja wrote: On 2015-05-18 22:10, Josh Berkus wrote: On 05/18/2015 01:04 PM, Ryan Pedela wrote: In the context of splitting shallow and deep merge into two operators, I think + is better for shallow and || better for deep. The reason for + is because many programming languages have this behavior. If I see the below code in language I have never used before: objC = objA + objB My default assumption is that + performs a shallow merge. Like I said, I would rather there just be one operator. Thank you, that helps. Anyone else? If everyone thinks the operators mean different things, we could just not add any operators and only provide functions instead. My $0.02: I would expect || to be what I want to use to add something to an existing JSON document, no matter what the path of what I'm adding is. In other words, deep merge. I certainly wouldn't expect it to be shallow. If we get this wrong now, we'll be stuck with it forever. At a minimum I think we should use anything other than || until we can figure this out. That leaves || available for whichever case we decide on. BTW, if people are set on shallow merge being || then I'd suggest ||| as the deep merge operator. -- Jim Nasby, Data Architect, Blue Treble Consulting Data in Trouble? Get it in Treble! http://BlueTreble.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
Peter Geoghegan p...@heroku.com writes: I think I agree with you, though: We should change things so that the relcache gives indexes in something like the ordering that you outline, rather than in the current arbitrary (though consistent) OID order. I'm fairly sure that there are aspects of the code that rely on indexes being returned by RelationGetIndexList() in a stable order. While I doubt that has to be exactly increasing-OID-order, I'm quite concerned about allowing the order to depend on mutable aspects of the indexes, like names. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
[HACKERS] small typo
Hi, Attached is a small typo. -- Euler Taveira Timbira - http://www.timbira.com.br/ PostgreSQL: Consultoria, Desenvolvimento, Suporte 24x7 e Treinamento diff --git a/contrib/pg_buffercache/pg_buffercache_pages.c b/contrib/pg_buffercache/pg_buffercache_pages.c index 98016fc..761c277 100644 --- a/contrib/pg_buffercache/pg_buffercache_pages.c +++ b/contrib/pg_buffercache/pg_buffercache_pages.c @@ -142,7 +142,7 @@ pg_buffercache_pages(PG_FUNCTION_ARGS) LWLockAcquire(BufMappingPartitionLockByIndex(i), LW_SHARED); /* - * Scan though all the buffers, saving the relevant fields in the + * Scan through all the buffers, saving the relevant fields in the * fctx-record structure. */ for (i = 0; i NBuffers; i++) -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] upper planner path-ification
At Tue, 19 May 2015 09:04:00 -0400, Robert Haas robertmh...@gmail.com wrote in CA+TgmobAV3_DS1sXA+PFWkjvX1K-VNiAnMMJrzPfD43g=-4...@mail.gmail.com On Tue, May 19, 2015 at 7:19 AM, Andrew Gierth and...@tao11.riddles.org.uk wrote: Tom == Tom Lane t...@sss.pgh.pa.us writes: Tom Hm. That's a hangover from when query_planner also gave back a Tom Plan (singular) rather than a set of Paths. I don't see any Tom fundamental reason why we couldn't generalize it to be a list of Tom potentially useful output orderings rather than just one. But I'm Tom a bit concerned about the ensuing growth in planning time; is it Tom really all that useful? The planning time growth is a possible concern, yes. The potential gain is eliminating one sort step, in the case when the input has a usable sorted path but grouping_planner happens not to ask for it (when there's more than just a single rollup, the code currently asks for one of the sort orders pretty much arbitrarily since it has no real way to know otherwise). Whether that would justify it... I don't know. Maybe that's one to save for later to see if there's any feedback from actual use. I kind of doubt that the growth in planning time would be anything too unreasonable. We already consider multiple orderings for ordinary base relations, so it's not very obvious why consideration multiple orderings for subqueries would be any worse. If we can arrange to throw away useless orderings early, as we do in other cases, then any extra paths we consider have a reasonable chance of being useful. Though I don't think that the simple path-ification of what is currently done make it grow in any degree, it could rapidly grow if we unconditionally construct extra upper-paths using the previously-abandoned extra paths or make them involved in join considerations. But the growth in planning time could be kept reasonable if we pay attention so that, as we have done so, the additional optimization schemes to have simple and precise bailing-out logic even if they require some complicated calculation or yields more extra paths. regards, -- Kyotaro Horiguchi NTT Open Source Software Center -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Change pg_cancel_*() to ignore current backend
Em terça-feira, 19 de maio de 2015, Jim Nasby jim.na...@bluetreble.com escreveu: On 5/19/15 6:30 PM, David G. Johnston wrote: On Tue, May 19, 2015 at 4:23 PM, Marko Tiikkaja ma...@joh.to mailto:ma...@joh.towrote: On 2015-05-20 00:59, Jim Nasby wrote: I find it annoying to have to specifically exclude pg_backend_pid() from pg_stat_activity if I'm trying to kill a bunch of backends at once, and I can't think of any reason why you'd ever want to call a pg_cancel_* function with your own PID. That's a rather easy way of testing that you're handling FATAL errors correctly from a driver/whatever. I'm having trouble thinking of a PC name for the function we create that should do this; while changing the pg_cancel_* functions to operate more safely. We could add a second parameter to the current functions: allow_own_pid DEFAULT false. To me that seems better than an entirely separate set of functions. +1 to add a second parameter to current functions. -- Fabrízio de Royes Mello Consultoria/Coaching PostgreSQL Timbira: http://www.timbira.com.br Blog: http://fabriziomello.github.io Linkedin: http://br.linkedin.com/in/fabriziomello Twitter: http://twitter.com/fabriziomello Github: http://github.com/fabriziomello
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On May 19, 2015 09:31:32 PM Greg Sabino Mullane wrote: Jan de Visser wrote: Well, one could argue that it *is* their problem, as they should be using the standard Postgres way for placeholders, which is $1, $2, $3... Shirley you are joking: Many products use JDBC as an abstraction layer facilitating (mostly) seamless switching between databases. I know the product I worked on did. Are you advocating that every single statement should use SELECT * FROM foo WHERE bar = $1 on pg and SELECT * FROM foo WHERE bar = ? on every other database? I'm not joking, and don't call me Shirley. If you are running into situations where you have question mark operators in your queries, you have already lost the query abstraction battle. There will be no seamless switching if you are using jsonb, hstore, ltree, etc. My statement was more about pointing out that Postgres already offers a complete placeholder system, which drivers are free to implement if they want. I must have misunderstood you strikeShirley/strike Greg, because to me it parsed as if you were suggesting (paraphrasing) ah forget about those pesky standardized drivers and their pesky syntax requirements. Just use ours like a big boy. I understand that once you start using '?' as (part of) operator names in your queries you're not portable anymore. I just thought that your proposed solution was to throw all portability out the window. But I was probably (hopefully?) wrong. jan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Back-branch update releases planned for next week
I wrote: In view of the multixactid wraparound issues that have gotten fixed over the last week or two, it's time to put out some minor releases. After some discussion among core and the packagers list, we concluded that we should do it next week (before the Memorial Day holiday). As per usual timing, we'll wrap tarballs Monday the 18th for public announcement Thursday the 21st. We encountered some, er, unplanned difficulties, which forced a re-wrap of the tarballs for 9.4.2 et al. To allow the packagers their usual amount of time to build packages, this week's releases will be announced on Friday the 22nd, not Thursday as would be the normal schedule. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Parallel Seq Scan
On Mon, May 11, 2015 at 3:00 AM, Robert Haas robertmh...@gmail.com wrote: I think it might be better to try to solve this problem in a more localized way. Can we arrange for planstate-instrumentation to point directory into the DSM, instead of copying the data over later? Yes, we can do that but I am not sure we can do that for pgBufferUsage which is a separate information we need to pass back to master backend. One way could be to change pgBufferUsage to a pointer and then allocate the memory for same at backend startup time and for parallel workers, it should point to DSM. Do you see any simple way to handle it? Another way could be that master backend waits for parallel workers to finish before collecting the instrumentation information and buffer usage stats. It seems to me that we need this information (stats) after execution in master backend is over, so I think we can safely assume that it is okay to finish the execution of parallel workers if they are not already finished the execution. With Regards, Amit Kapila. EnterpriseDB: http://www.enterprisedb.com
[HACKERS] Per row status during INSERT .. ON CONFLICT UPDATE?
Hi, Is there a way to know which rows were INSERTed and UPDATEd when doing a INSERT ... ON CONFLICT UPDATE? Probably via pseudo column indicating INSERT / UPDATE ? The RETURNING clause just allows us to return columns, but am unable to find a way to know 'what' happened to a given row. Any pointers would be helpful. Couldn't find anything related in 9.5devel docs either. -- thanks Robins
Re: [HACKERS] Run pgindent now?
On 05/18/2015 08:06 PM, Andrew Dunstan wrote: On 05/18/2015 07:04 PM, Bruce Momjian wrote: On Mon, May 18, 2015 at 06:53:00PM -0400, Tom Lane wrote: Robert Haas robertmh...@gmail.com writes: On Mon, May 18, 2015 at 12:10 PM, Bruce Momjian br...@momjian.us wrote: There was talk last time of pgindent-ing head and all back branches, because a patch applied to head and back branches was historically only pgindented in head, meaning that any future patches in that area could not be easily backpatched. Do we want to do this? I am personally not excited about that. I would rather leave the back-branches alone. It would be awfully nice though if we didn't have to deal with random cross-branch indenting differences. I've lost, maybe not years off my life, but certainly weeks of not-very-pleasant make-work because of that. I'm surprised you've not had the same experience. If people were good about pgindenting patches meant to be back-patched *before* they committed, it would not be such an issue, but they're not very good about that. I couldn't figure out why we were getting that code drift, but now that Tom has identified why it happens, it seems good that we fix it. Would it alleviate your concern any if we eased into this, like say only apply the back-branch pgindent run to 9.5 and later branches? Then at least I could foresee the end of that particular annoyance. (BTW, one practical issue is where would we get typedef lists relevant to the back branches. I'm not sure if the buildfarm infrastructure is capable of collecting branch-specific data, or if we'd need to rather than just using a union of all branches' typedefs.) Uh, I just happen to commit the typedef list file used for the pgindent run in src/tools/pgindent/typedefs.list, per branch, so we would just use the same file. If typedefs were added in a backbranch (unlikely), we probably wouldn't want to use them anyway. The buildfarm animals are perfectly capable of finding typedefs for each branch. They haven't been because the default configuration is only to collect them for HEAD. Changing this is easy, especially since I control five of the six members currently reporting typedefs successfully, and Tom controls the other one. I've currently set two of them to do run typedefs for all live branches. The other thing is that the server script that amalgamates them only looks at HEAD. That will need to change. We would probably want an amalgamated list, because there could have been symbols on old branches that were deleted in later branches. With luck the presence of false positives wouldn't matter. It usually doesn't seem to. OK, if you look at http://www.pgbuildfarm.org/cgi-bin/typedefs.pl?show_list you will be able to see the state of things. It's not even remotely pretty, and I am going to fix that, but it works. As you will be able to see, a number of buildfarm members are reporting on typedefs on all the live branches. You can get the list for each branch by hitting the appropriate link (essentially '/cgi-bin/typedefs.pl?branch=$branch'). If you ask for 'ALL' as the branch it gives you the amalgamated list over all branches. If you don't specify a branch at all, it gives you HEAD (which is buildfarm spelling for master), since that's what it did previously. I can change the default to ALL if that's what people want. Tom, if you want to get dromedary reporting on all branches, just remove the branches = [ 'HEAD' ], from the config. Enjoy. cheers andrew -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
David G. Johnston david.g.johns...@gmail.com wrote:
On Mon, May 18, 2015 at 3:31 PM, Bruno Harbulot
br...@distributedmatter.netwrote:
In the discussion on the OpenJDK JDBC list two years ago
(
http://mail.openjdk.java.net/pipermail/jdbc-spec-discuss/2013-February/50.html
),
Lance Andersen said There is nothing in the SQL standard that
would support the use of an '?' as anything but a parameter
marker..
CREATE OPERATOR is a PostgreSQL extension. There are no
provisions for user-defined operators in the SQL standard.
Exactly. The standard specifies the characters to use for the
predicates that it defines, and provides no mechanism for adding
additional predicates; but who in the world would want to exclude
all extensions to the standard?
And by extension if indeed the standard does require the use of
? for parameters we are in violation there because the backend
protocol deals with $# placeholders and not ?
We're talking about a different specification that has question
marks as parameter placeholders. That's in the Java Database
Connector (JDBC) specification. (It is apparently also specified
in other documents, although I'm not familiar enough with those to
comment.) Note that it would create all sorts of pain if both the
SQL statements and a connector issuing them used the same
convention for substituting parameters; it is a *good* thing that
plpgsql and SQL function definitions use a different convention
than JDBC!
The JDBC spec provides for escapes using curly braces (including
product-specific escapes); it seems like a big mistake for us to
have chosen a completely different mechanism for escaping the
question mark character in a SQL statement. Perhaps the least
painful path would be to add support for {?} as the escape for a
question mark, and a connection option to supplement that with
support for the legacy \? escape. I would bet a lot of money that
even with an if test for that option, the curly brace escape
would be faster than what's there now (when the option was not
set). Some operators would look a little funny in Java string
literals, but that's not so bad.
--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Run pgindent now?
On Mon, May 18, 2015 at 6:53 PM, Tom Lane t...@sss.pgh.pa.us wrote: I am personally not excited about that. I would rather leave the back-branches alone. It would be awfully nice though if we didn't have to deal with random cross-branch indenting differences. I've lost, maybe not years off my life, but certainly weeks of not-very-pleasant make-work because of that. I'm surprised you've not had the same experience. Well, there are a couple of things that worry me: - People rely on us to ship, in minor releases, only critical security and stability fixes. Re-indenting the code is neither, and people may not appreciate needless whitespace differences being shipped in the next branch. Anyone who diffs that tarball against the previous one is going to see a bunch of stuff in there that may make them nervous. - If pgindent doesn't handle every branch in exactly the same way, it's possible that this change could exacerbate differences instead of reducing them. I actually think this is quite a likely outcome. I personally have not found back-patching to have been significantly complicated by whitespace differences. There are certainly code differences that can make it quite miserable in some cases, but I cannot recall a case where there was an issue of this time due to erratic indenting in one branch that had meanwhile been fixed in another branch. I accept that your experience may be different, of course. Would it alleviate your concern any if we eased into this, like say only apply the back-branch pgindent run to 9.5 and later branches? Then at least I could foresee the end of that particular annoyance. If we do this only beginning with 9.5, and if we can make the output 100% consistent across branches, and if we run it before EVERY minor release so that people don't see unrelated diffs between consecutive tarballs, then it would address my concerns. I wish that pgident could be made more automated, like by having it fully built into the tree so that you can type 'make indent', or by having a daemon that would automatically pgindent the main tree periodically (say, once a month, or when more than X number of lines/files have changed, whichever comes first). I still find it quite a hassle to set up and run. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Per row status during INSERT .. ON CONFLICT UPDATE?
On 5/19/15 3:04 PM, Thom Brown wrote:
If you want the delta, you'll have to resort to a CTE:
e.g.
# WITH newvals AS (
INSERT INTO test (name, age) VALUES ('James', 45)
ON CONFLICT (name)
DO UPDATE SET age = EXCLUDED.age
RETURNING *)
SELECT n.name, o.age as old.age, n.age as new.age
FROM test o RIGHT JOIN newvals n on o.name = n.name;
name | old.age | new.age
---+-+-
James | 44 | 45
(1 row)
Also note that the old value is not the actual value right before the
update, but one according to a snapshot taken at the beginning of the
query. So if you instead did SET age = age + 1, you could see an old
value of 44 and a new value of 46 (or any similarly weird combination of
values).
.m
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Patch for bug #12845 (GB18030 encoding)
That's fine when not every code point is used, but it's different for
GB18030 where almost all code points are used. Using a plain array
saves space and saves a binary search.
Well, it doesn't save any space: if we get rid of the additional linear
ranges in the lookup table, what remains is 30733 entries requiring about
256K, same as (or a bit less than) what you suggest.
We could do both. What about something like this:
static unsigned int utf32_to_gb18030_from_0x0001[1105] = {
/* 0x0 */ 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8,
...
static unsigned int utf32_to_gb18030_from_0x2010[1587] = {
/* 0x0 */ 0xa95c, 0x8136a532, 0x8136a533, 0xa843, 0xa1aa, 0xa844,
0xa1ac, 0x8136a534,
...
static unsigned int utf32_to_gb18030_from_0x2E81[28965] = {
/* 0x0 */ 0xfe50, 0x8138fd39, 0x8138fe30, 0xfe54, 0x8138fe31,
0x8138fe32, 0x8138fe33, 0xfe57,
...
static unsigned int utf32_to_gb18030_from_0xE000[2149] = {
/* 0x0 */ 0xaaa1, 0xaaa2, 0xaaa3, 0xaaa4, 0xaaa5, 0xaaa6, 0xaaa7, 0xaaa8,
...
static unsigned int utf32_to_gb18030_from_0xF92C[254] = {
/* 0x0 */ 0xfd9c, 0x84308535, 0x84308536, 0x84308537, 0x84308538,
0x84308539, 0x84308630, 0x84308631,
...
static unsigned int utf32_to_gb18030_from_0xFE30[464] = {
/* 0x0 */ 0xa955, 0xa6f2, 0x84318538, 0xa6f4, 0xa6f5, 0xa6e0, 0xa6e1, 0xa6f0,
...
static uint32
conv_utf8_to_18030(uint32 code)
{
uint32 ucs = utf8word_to_unicode(code);
#define conv_lin(minunicode, maxunicode, mincode) \
if (ucs = minunicode ucs = maxunicode) \
return gb_unlinear(ucs - minunicode + gb_linear(mincode))
#define conv_array(minunicode, maxunicode) \
if (ucs = minunicode ucs = maxunicode) \
return utf32_to_gb18030_from_##minunicode[ucs - minunicode];
conv_array(0x0001, 0x0452);
conv_lin(0x0452, 0x200F, 0x8130D330);
conv_array(0x2010, 0x2643);
conv_lin(0x2643, 0x2E80, 0x8137A839);
conv_array(0x2E81, 0x9FA6);
conv_lin(0x9FA6, 0xD7FF, 0x82358F33);
conv_array(0xE000, 0xE865);
conv_lin(0xE865, 0xF92B, 0x8336D030);
conv_array(0xF92C, 0xFA2A);
conv_lin(0xFA2A, 0xFE2F, 0x84309C38);
conv_array(0xFE30, 0x1);
conv_lin(0x1, 0x10, 0x90308130);
/* No mapping exists */
return 0;
}
The point about possibly being able to do this with a simple lookup table
instead of binary search is valid, but I still say it's a mistake to
suppose that we should consider that only for GB18030. With the reduced
table size, the GB18030 conversion tables are not all that far out of line
with the other Far Eastern conversions:
$ size utf8*.so | sort -n
textdata bss dec hex filename
1880 512 162408 968 utf8_and_ascii.so
2394 528 162938 b7a utf8_and_iso8859_1.so
6674 512 1672021c22 utf8_and_cyrillic.so
24318 904 16 252386296 utf8_and_win.so
28750 968 16 297347426 utf8_and_iso8859.so
121110 512 16 121638 1db26 utf8_and_euc_cn.so
123458 512 16 123986 1e452 utf8_and_sjis.so
133606 512 16 134134 20bf6 utf8_and_euc_kr.so
185014 512 16 185542 2d4c6 utf8_and_sjis2004.so
185522 512 16 186050 2d6c2 utf8_and_euc2004.so
212950 512 16 213478 341e6 utf8_and_euc_jp.so
221394 512 16 221922 362e2 utf8_and_big5.so
274772 512 16 275300 43364 utf8_and_johab.so
26 512 16 278304 43f20 utf8_and_uhc.so
332262 512 16 332790 513f6 utf8_and_euc_tw.so
350640 512 16 351168 55bc0 utf8_and_gbk.so
496680 512 16 497208 79638 utf8_and_gb18030.so
If we were to get excited about reducing the conversion time for GB18030,
it would clearly make sense to use similar infrastructure for GBK, and
perhaps the EUC encodings too.
I'll check them as well. If they have linear ranges it should work.
However, I'm not that excited about changing it. We have not heard field
complaints about these converters being too slow. What's more, there
doesn't seem to be any practical way to apply the same idea to the other
conversion direction, which means if you do feel there's a speed problem
this would only halfway fix it.
It does work if you linearlize it first. That's why we need to convert
to utf32 first as well. That's a form of linearization.
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On 18 May 2015 at 18:49, David G. Johnston david.g.johns...@gmail.com wrote: On Mon, May 18, 2015 at 3:31 PM, Bruno Harbulot br...@distributedmatter.net wrote: On Sun, May 17, 2015 at 5:15 PM, Greg Sabino Mullane g...@turnstep.com wrote: In that case my vote is new operators. This has been a sore point for the JDBC driver Um, no, new operators is a bad idea. Question marks are used by hstore, json, geometry, and who knows what else. I think the onus is solely on JDBC to solve this problem. DBD::Pg solved it in 2008 with the pg_placeholder_dollaronly solution, and earlier this year by allowing backslashes before the question mark (because other parts of the stack were not able to smoothly implement pg_placeholder_dollaronly.) I recommend all drivers implement \? as a semi-standard workaround. See also: http://blog.endpoint.com/2015/01/dbdpg-escaping-placeholders-with.html I'm not sure the onus is solely on JDBC. Using question marks in operators clearly has required a number of connectors to implement their own workarounds, in different ways. This also seems to affect some libraries and frameworks that depend on those connectors (and for which the workarounds may even be more convoluted). My main point was that this is not specific to JDBC. Considering that even PostgreSQL's own ECPG is affected, the issue goes probably deeper than it seems. I'm just not convinced that passing the problem onto connectors, libraries and ultimately application developers is the right thing to do here. In the discussion on the OpenJDK JDBC list two years ago ( http://mail.openjdk.java.net/pipermail/jdbc-spec-discuss/2013-February/50.html ), Lance Andersen said There is nothing in the SQL standard that would support the use of an '?' as anything but a parameter marker.. It might be worth finding out whether this is indeed the case according to the SQL specifications (I'm afraid I'm not familiar with these specifications to do it myself). CREATE OPERATOR is a PostgreSQL extension. There are no provisions for user-defined operators in the SQL standard. http://www.postgresql.org/docs/9.4/interactive/sql-createoperator.html And by extension if indeed the standard does require the use of ? for parameters we are in violation there because the backend protocol deals with $# placeholders and not ? I too do not know enough here. Note that it would not be enough to change the existing operators - any use of ? would have to be forbidden including those created by users. The first step on this path would be for someone to propose a patch adding alternative operators for every existing operator that uses ?. If this idea is to move forward at all that patch would have to be accepted. Such a patch is likely to see considerable bike-shedding. We then at least provide an official way to avoid ? operators that shops can make use of at their discretion. Removing the existing operators or forbidding custom operators is a separate discussion. David J. It would seem that choosing ? for operators was ill advised; I'm not convinced that deprecating them is a bad idea. If we start now, in 5 years they should be all but gone Agreed a patch would be the first place to start Dave Cramer dave.cramer(at)credativ(dot)ca http://www.credativ.ca
Re: [HACKERS] Per row status during INSERT .. ON CONFLICT UPDATE?
On 19 May 2015 at 13:23, Robins Tharakan thara...@gmail.com wrote:
Hi,
Is there a way to know which rows were INSERTed and UPDATEd when doing a
INSERT ... ON CONFLICT UPDATE? Probably via pseudo column indicating INSERT
/ UPDATE ?
The RETURNING clause just allows us to return columns, but am unable to find
a way to know 'what' happened to a given row.
Any pointers would be helpful.
Couldn't find anything related in 9.5devel docs either.
I don't think there's anything that tells you directly in the results
whether an INSERT or an UPDATE was performed. But you could use a
hack which is to return the xmax in the output, and if that's 0, it
INSERTed. If it's greater than 0, it UPDATEd:
e.g.
# INSERT INTO test (name, age) values ('Jack', 44) ON CONFLICT (name)
DO UPDATE SET age = EXCLUDED.age RETURNING xmax, *;
xmax | id | name | age
--++--+-
0 | 70 | Jack | 44
(1 row)
# INSERT INTO test (name, age) values ('Jack', 44) ON CONFLICT (name)
DO UPDATE SET age = EXCLUDED.age RETURNING xmax, *;
xmax | id | name | age
-++--+-
1097247 | 70 | Jack | 44
(1 row)
If you want the delta, you'll have to resort to a CTE:
e.g.
# WITH newvals AS (
INSERT INTO test (name, age) VALUES ('James', 45)
ON CONFLICT (name)
DO UPDATE SET age = EXCLUDED.age
RETURNING *)
SELECT n.name, o.age as old.age, n.age as new.age
FROM test o RIGHT JOIN newvals n on o.name = n.name;
name | old.age | new.age
---+-+-
James | 44 | 45
(1 row)
Regards
Thom
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] a few thoughts on the schedule
On Mon, May 18, 2015 at 11:52 PM, Andres Freund and...@anarazel.de wrote: [first 9.6 CF around 2015-07-15] Honestly, that seems awful soon. I would have thought maybe August 15th. Maybe we should just rename it to 9.6-1 for now? And then look how things look around pgcon? I'd rather agree on a date. People need to plan their schedules. I am inclined to think the 5-CommitFest thing we did this time did not work out. It might've been fine if feature freeze had been a month earlier, but by freezing in May we've pretty clearly stolen at least a month, if not two, from the next cycle. I personally think the late close of the 9.4 cycle has alone thrings far enough off track that we can't fairly evaluate a 5 CF schedule. Oh, I agree with that. I'm certainly not saying we shouldn't do it again. But I don't see a practical way to do 5 CFs again for 9.6 and also release it in September of 2016. I don't think it would be a good idea to open the tree for 9.6 development in three weeks, or even in time for a July 1st CommitFest. The vary earliest time frame that would make sense to me is to branch July 1st and start a CF on July 15th. If we schedule four more CommitFests after that at two month intervals, they would start on September 15th, November 15th, January 15th, and March 15th, putting us a month behind where we were this time. That's not going to work. So I think the options are: - Do 4 CommitFests as we have for past releases. We could do July 15th, September 15th, November 15th, and January 15th; or we could do August 1st, October 1st, December 1st, and February 1st; or we could do August 15th, October 15th, December 15th, and February 15th. Probably, that last one isn't so good: starting on December 15th is going to suck. - Do 5 or more CommitFests and accept that the release cycle is going to be more than a year. Personally, given where we're at right now, I don't think an early fall release of 9.5 is going to be remotely practical. I think we're going to end up releasing in late fall or early in the new year. So I'd be completely fine with a schedule that aims for 9.6 to get released around March-May of 2017, so the last CommitFest would start in August or September of 2016. I expect that to be unpopular, which is fine, but then I think we have to limit ourselves to 4 CFs this time through. Personally I'm coming more and more to the conclusion that CFs just don't work [anymore]. I think the *tracking* itself is rather important and has a worthwhile role. But it seems to me that what CFs have lately essentially ended up being, is closer to a cycle long review queue than anything else. I mostly agree with that. ISTM that the CF scheduling right now does more harm than good. * They seem to frustrate a lot of the people doing a lot of reviews. * Evidently they don't very well prevent individual patches from just slipping through and through. * They lead to completely uninteresting patches being reviewed before others. * The contribution experience is still pretty painful and takes ages Those are legitimate issues. Maybe we should forget them and just have monthly 'judgefests' where some poor sod summarizes the current state and direction, and we then collaboratively discuss whether we see things going anywhere and if not, what would need to happen that they do. And have a policy that older patches should be preferred over newer ones; but at the same time cull patches continually sitting at the tail end as 'not interesting'. I think we need to start by understanding that we need the contribution experience to be good for both patch authors and also for reviewers (including reviewers who are commiters). We very much need to give new contributors a good experience of submitting patches and getting useful feedback and getting stuff committed. I think it's clear that we could do a much better job at that, and the project would benefit enormously. However, doing a better job means spending more time on it, and we can't just demand that senior reviewers or contributors spend more time on it. I mean, we can, I guess, but it will only breed frustration and resentment. I'm not sure what the solution is here, but if it boils down to telling people who have put a lot of effort into the project over a long period of time that they are not doing enough, I'm here to say that won't work. So one problem that comes up in the context of your proposal is that it's likely to be hard to find the poor sod whose existence you hypothecate. Maybe there is someone who will do that once or twice, but I think it'll be hard to keep that position filled over the long term. Unfortunately, I don't have a lot of good ideas here. I know that I spend as much time reviewing other people's patches as I can manage to find in my schedule, and I know a lot of people would probably like to see me do more of that. I'm sure there are also some people who would like to see me do
Re: [HACKERS] Run pgindent now?
Robert Haas robertmh...@gmail.com writes: On Mon, May 18, 2015 at 6:53 PM, Tom Lane t...@sss.pgh.pa.us wrote: Would it alleviate your concern any if we eased into this, like say only apply the back-branch pgindent run to 9.5 and later branches? Then at least I could foresee the end of that particular annoyance. If we do this only beginning with 9.5, and if we can make the output 100% consistent across branches, and if we run it before EVERY minor release so that people don't see unrelated diffs between consecutive tarballs, then it would address my concerns. To do it before every minor release would require re-indenting HEAD as well (since the whole point is to keep HEAD and the back branches consistent). I think we'd get too much push-back from developers whose pending patches got broken. We can get away with reindenting HEAD between development cycles, but probably not more often than that. I'm not particularly concerned by the tarball-diff argument: running diff with --ignore-spaces should mask most of the changes. Moreover, assuming the code was properly indented at x.y.0 release time, any changes applied by pgindent would only be within subsequent back-patches, which hopefully are a very small part of the code. (Perhaps it would be useful to do a trial indent on some old branch right now, just to see how large the diffs are; then we'd have some actual facts in this argument...) And lastly, committers who are bothered by the prospect of such changes could take the time to reindent their back-patched changes before committing in the first place. (FWIW, I usually do, and it's not hard except in files that have been heavily mangled in HEAD.) I wish that pgident could be made more automated, like by having it fully built into the tree so that you can type 'make indent', or by having a daemon that would automatically pgindent the main tree periodically (say, once a month, or when more than X number of lines/files have changed, whichever comes first). I still find it quite a hassle to set up and run. It is a pain. I have a shell script that fetches the typedef list automatically, which helps. The main problem with a make indent target is that only in Bruce's annual runs do we really want to let it loose on the whole tree. In manual fixups, I only point it at the files I've edited (and then, often, I have to remove some diffs in unrelated parts of those files). I wish that could be a bit easier, though I'm not sure how. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On 19 May 2015 at 10:23, Kevin Grittner kgri...@ymail.com wrote:
David G. Johnston david.g.johns...@gmail.com wrote:
On Mon, May 18, 2015 at 3:31 PM, Bruno Harbulot
br...@distributedmatter.netwrote:
In the discussion on the OpenJDK JDBC list two years ago
(
http://mail.openjdk.java.net/pipermail/jdbc-spec-discuss/2013-February/50.html
),
Lance Andersen said There is nothing in the SQL standard that
would support the use of an '?' as anything but a parameter
marker..
CREATE OPERATOR is a PostgreSQL extension. There are no
provisions for user-defined operators in the SQL standard.
Exactly. The standard specifies the characters to use for the
predicates that it defines, and provides no mechanism for adding
additional predicates; but who in the world would want to exclude
all extensions to the standard?
And by extension if indeed the standard does require the use of
? for parameters we are in violation there because the backend
protocol deals with $# placeholders and not ?
We're talking about a different specification that has question
marks as parameter placeholders. That's in the Java Database
Connector (JDBC) specification. (It is apparently also specified
in other documents, although I'm not familiar enough with those to
comment.) Note that it would create all sorts of pain if both the
SQL statements and a connector issuing them used the same
convention for substituting parameters; it is a *good* thing that
plpgsql and SQL function definitions use a different convention
than JDBC!
The JDBC spec provides for escapes using curly braces (including
product-specific escapes); it seems like a big mistake for us to
have chosen a completely different mechanism for escaping the
question mark character in a SQL statement. Perhaps the least
painful path would be to add support for {?} as the escape for a
question mark, and a connection option to supplement that with
support for the legacy \? escape. I would bet a lot of money that
even with an if test for that option, the curly brace escape
would be faster than what's there now (when the option was not
set). Some operators would look a little funny in Java string
literals, but that's not so bad.
Perhaps reviewing https://github.com/pgjdbc/pgjdbc/pull/187 might help
understand why we chose ??
Dave Cramer
dave.cramer(at)credativ(dot)ca
http://www.credativ.ca
[HACKERS] Wrong Assert in PageIndexMultiDelete?
Hi, hackers! I am trying to create new index access method. And I found strange Assert in PageIndexMultiDelete http://doxygen.postgresql.org/bufpage_8c_source.html#l00791 function. Assert http://doxygen.postgresql.org/c_8h.html#a706ac5b1a53bd04067f81924b92cb9f6(nitems MaxIndexTuplesPerPage http://doxygen.postgresql.org/itup_8h.html#adb7c94e95ce112eb47d71f7797604ddb ); Is '' sign is correct? I thougt it should be '='. Is it a bug or just my misunderstanding? -- Best regards, Lubennikova Anastasia
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
On Mon, May 18, 2015 at 12:33 PM, Alvaro Herrera alvhe...@2ndquadrant.com wrote: Bruce Momjian wrote: On Sun, May 17, 2015 at 09:31:47PM +0200, José Luis Tallón wrote: On 05/17/2015 07:39 PM, Tom Lane wrote: =?windows-1252?Q?Jos=E9_Luis_Tall=F3n?= jltal...@adv-solutions.net writes: On the other hand, ISTM that what we all intend to achieve is some Postgres equivalent of the SUID bit... so why not just do something equivalent? --- LOGIN-- as user with the appropriate role membership / privilege? ... SET ROLE / SET SESSION AUTHORIZATION WITH COOKIE / IMPERSONATE ... do whatever ...-- unprivileged user can NOT do the impersonate thing DISCARD ALL-- implicitly restore previous authz --- Oh? What stops the unprivileged user from doing DISCARD ALL? Indeed. The pooler would need to block this. Or we would need to invent another (this time, privileged) verb in order to restore authz. What if you put the SQL in a function then call the function? I don't see how the pooler could block this. I think the idea of having SET SESSION AUTH pass a cookie, and only let RESET SESSION AUTH work when the same cookie is supplied, is pretty reasonable. That seems like a kludge to me. If the cookie leaks out somhow, which it will, then it'll be insecure. I think the way to do this is with a protocol extension that poolers can enable on request. Then they can just refuse to forward any reset authorization packets they get from their client. There's no backward-compatibility break because the pooler can know, from the server version, whether the server is new enough to support the new protocol messages. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On Tue, May 19, 2015 at 3:23 PM, Kevin Grittner kgri...@ymail.com wrote: David G. Johnston david.g.johns...@gmail.com wrote: On Mon, May 18, 2015 at 3:31 PM, Bruno Harbulot br...@distributedmatter.netwrote: In the discussion on the OpenJDK JDBC list two years ago ( http://mail.openjdk.java.net/pipermail/jdbc-spec-discuss/2013-February/50.html ), Lance Andersen said There is nothing in the SQL standard that would support the use of an '?' as anything but a parameter marker.. CREATE OPERATOR is a PostgreSQL extension. There are no provisions for user-defined operators in the SQL standard. Exactly. The standard specifies the characters to use for the predicates that it defines, and provides no mechanism for adding additional predicates; but who in the world would want to exclude all extensions to the standard? I was certainly not suggesting custom operators should be excluded. I was suggesting using something that was actually not incompatible with the SQL standards (and, even with standards aside, the expectations implementors have regarding the question mark, since it affects other tools too). And by extension if indeed the standard does require the use of ? for parameters we are in violation there because the backend protocol deals with $# placeholders and not ? We're talking about a different specification that has question marks as parameter placeholders. That's in the Java Database Connector (JDBC) specification. (It is apparently also specified in other documents, although I'm not familiar enough with those to comment.) Note that it would create all sorts of pain if both the SQL statements and a connector issuing them used the same convention for substituting parameters; it is a *good* thing that plpgsql and SQL function definitions use a different convention than JDBC! Actually, we were not just talking about JDBC. I don't know the specifications in details, but the SQL:201x (preliminary) documents linked from https://wiki.postgresql.org/wiki/Developer_FAQ#Where_can_I_get_a_copy_of_the_SQL_standards.3F seem to have some information. The Foundation document (Section 4.25 Dynamic SQL concepts) says that dynamic parameters are represented by a question mark. In addition, the BNF grammar available at http://www.savage.net.au/SQL/sql-2003-2.bnf.html#dynamic%20parameter%20specification also says: dynamic parameter specification::= question mark I'm not familiar enough with these documents to know whether I'm missing some context, but it would seem that the question mark is a reserved character, beyond the scope of JDBC (at the very least, it seems incompatible with Dynamic SQL and its implementation in ECPG). Best wishes, Bruno.
Re: [HACKERS] a few thoughts on the schedule
On 05/18/2015 08:52 PM, Andres Freund wrote: Maybe we should forget them and just have monthly 'judgefests' where some poor sod summarizes the current state and direction, and we then collaboratively discuss whether we see things going anywhere and if not, what would need to happen that they do. And have a policy that older patches should be preferred over newer ones; but at the same time cull patches continually sitting at the tail end as 'not interesting'. I don't think this will be a productive solution. I would argue that any solution we come up with, somebody is going to think they got the short end of the stick. There will be someone that thinks it is inefficient, that it doesn't suit their needs or that it doesn't work in their paradigm. That is why we don't have a proper issue/bug tracker. That is why we are constantly inventing here instead of relying on the work of others (when it comes to this particular problem). I don't know what the solution is but I know I like the idea of a tree freeze except for bug fixes for at least 3 weeks but I would be jumping for joy if we froze the tree except for bug fixes for 6 or 12 weeks. I don't care about 9.6 at this point. We move so fast anyway, most people I know haven't even migrated to 9.4.x and even more are happily plugging away on 9.2. Consider that yes, we have a ton of people that migrated to 9.4 but those generally aren't people running the 24x7 enterprise class database. It will not hurt us, and will only help us to slow down for this release. If 9.6 gets pushed until Winter 2017, so what. Let's release Alpha1, start promoting the heck out of it amongst the community and early adopting (for NON PRODUCTION) developers. Let's make it easy as snot dripping from a toddlers nose to submit bug reports. Let's verify those things and let's produce the most solid, reliable and bug free PostgreSQL version, ever. The summer is nigh and it is going to be slow going anyway. JD -- Command Prompt, Inc. - http://www.commandprompt.com/ 503-667-4564 PostgreSQL Centered full stack support, consulting and development. Announcing I'm offended is basically telling the world you can't control your own emotions, so everyone else should do it for you. -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
Mike Blackwell mike.blackw...@rrd.com writes: See for example http://docs.oracle.com/cd/B19306_01/text.102/b14218/cqoper.htm#i997330, Table 3-1, third row, showing the precedence of '?'. Further down the page, under Fuzzy see Backward Compatibility Syntax. If I'm reading that right, that isn't a SQL-level operator but an operator in their text search query language, which would only appear in SQL queries within string literals (compare tsquery's query operators in PG). So it wouldn't be a hazard for ?-substitution, as long as the substituter was bright enough to not change string literals. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] a few thoughts on the schedule
On 05/19/2015 11:02 AM, Peter Geoghegan wrote: Hasn't every talented reviewer gotten job offers shortly afterwards in the last few years? The ones that accept don't necessarily work that much in the community, but several seem to. And I think in the case of several people the reason they don't, is less the company, but that it's emotionally draining. I think that's very true, and often unacknowledged. Reviewing other people's work can be very difficult. I do not enjoy conflict with other people on this mailing list one bit, and that's getting harder to deal with on a personal level over time, not easier. Although I certainly understand your sentiment. It isn't personal. It is professional. If people are taking personally (and I certainly have), they need to step the heck back, take a breath and ask themselves what their problem is. If they won't, then someone in the community needs to step up and help them do that. The more we remove the ID, the more productive we will become. Sincerely, JD -- Command Prompt, Inc. - http://www.commandprompt.com/ 503-667-4564 PostgreSQL Centered full stack support, consulting and development. Announcing I'm offended is basically telling the world you can't control your own emotions, so everyone else should do it for you. -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
On 2015-05-19 14:41:06 -0400, Robert Haas wrote: On Tue, May 19, 2015 at 12:29 PM, Andres Freund and...@anarazel.de wrote: On 2015-05-19 10:53:10 -0400, Robert Haas wrote: That seems like a kludge to me. If the cookie leaks out somhow, which it will, then it'll be insecure. I think the way to do this is with a protocol extension that poolers can enable on request. Then they can just refuse to forward any reset authorization packets they get from their client. There's no backward-compatibility break because the pooler can know, from the server version, whether the server is new enough to support the new protocol messages. That sounds like a worse approach to me. Don't you just need to hide the session authorization bit in a function serverside to circumvent that? I'm apparently confused. There's nothing you can do to maintain security against someone who can load C code into the server. I must be misunderstanding you. It very well might be me that's confused. But what's stopping a user from doing a RESET SESSION AUTHORIZATION; in a DO block or something? I guess you are intending that a RESET SESSION AUTHORIZATION is only allowed on a protocol level when the protocol extension is in use? Greetings, Andres Freund -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On 19 May 2015 at 16:32, I wrote: In the event that the INSERT triggers a constraint that the UPDATE fails to resolve, it will still fail in exactly the same way that running the ON CONFLICT on a specific constraint would fail, so it's not like you gain any extra value from specifying the constraint, is it? I don't know why I wrote this paragraph, it's just the product of me thinking of something else at the same time: UPDATE obviously doesn't resolve a conflict as such. Thinking about it more, I suppose if multiple constraints end up triggering for the same INSERT, it would require UPDATEs of multiple rows. Is that the issue? Geoff
Re: [HACKERS] errmsg() clobbers errno
John Gorman johngorm...@gmail.com writes: While debugging an extension I discovered that the errmsg() function zeros out errno. So might a lot of other functions used in an ereport's arguments. This is annoying because if the process of assembling a meaningful error message happens to call errmsg() before calling strerror() we lose the strerror information. This is why you should use %m and not strerror(errno). The infrastructure for %m is set up so that errno is captured before evaluating any of the ereport's arguments. I am attaching a patch to preserve errno across errmsg() calls. This is pretty useless, unfortunately, because there are just too many ways to bite yourself on the rear if you reference errno inside the arguments of an ereport (or any other complicated nest of function calls). regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
On 2015-05-19 10:53:10 -0400, Robert Haas wrote: That seems like a kludge to me. If the cookie leaks out somhow, which it will, then it'll be insecure. I think the way to do this is with a protocol extension that poolers can enable on request. Then they can just refuse to forward any reset authorization packets they get from their client. There's no backward-compatibility break because the pooler can know, from the server version, whether the server is new enough to support the new protocol messages. That sounds like a worse approach to me. Don't you just need to hide the session authorization bit in a function serverside to circumvent that? Greetings, Andres Freund -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] upper planner path-ification
On Tue, May 19, 2015 at 7:19 AM, Andrew Gierth and...@tao11.riddles.org.uk wrote: Tom == Tom Lane t...@sss.pgh.pa.us writes: Tom Hm. That's a hangover from when query_planner also gave back a Tom Plan (singular) rather than a set of Paths. I don't see any Tom fundamental reason why we couldn't generalize it to be a list of Tom potentially useful output orderings rather than just one. But I'm Tom a bit concerned about the ensuing growth in planning time; is it Tom really all that useful? The planning time growth is a possible concern, yes. The potential gain is eliminating one sort step, in the case when the input has a usable sorted path but grouping_planner happens not to ask for it (when there's more than just a single rollup, the code currently asks for one of the sort orders pretty much arbitrarily since it has no real way to know otherwise). Whether that would justify it... I don't know. Maybe that's one to save for later to see if there's any feedback from actual use. I kind of doubt that the growth in planning time would be anything too unreasonable. We already consider multiple orderings for ordinary base relations, so it's not very obvious why consideration multiple orderings for subqueries would be any worse. If we can arrange to throw away useless orderings early, as we do in other cases, then any extra paths we consider have a reasonable chance of being useful. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Run pgindent now?
Andrew Dunstan and...@dunslane.net writes: Tom, if you want to get dromedary reporting on all branches, just remove the branches = [ 'HEAD' ], from the config. dromedary is a pretty slow machine, so I'm going to pass on that unless there's a good reason to think it would find typedefs your machines don't. I rather doubt that --- our use of platform-dependent typedefs is fairly small and stable, so it seems like checking HEAD should be sufficient. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
A Google search suggests Oracle 9.x supports a unary '?' operator (fuzzy match), so the use of '?' in an operator name is not without precedent. __ *Mike Blackwell | Technical Analyst, Distribution Services/Rollout Management | RR Donnelley* 1750 Wallace Ave | St Charles, IL 60174-3401 Office: 630.313.7818 mike.blackw...@rrd.com http://www.rrdonnelley.com http://www.rrdonnelley.com/ * mike.blackw...@rrd.com* On Tue, May 19, 2015 at 10:03 AM, Bruno Harbulot br...@distributedmatter.net wrote: On Tue, May 19, 2015 at 3:23 PM, Kevin Grittner kgri...@ymail.com wrote: David G. Johnston david.g.johns...@gmail.com wrote: On Mon, May 18, 2015 at 3:31 PM, Bruno Harbulot br...@distributedmatter.netwrote: In the discussion on the OpenJDK JDBC list two years ago ( http://mail.openjdk.java.net/pipermail/jdbc-spec-discuss/2013-February/50.html ), Lance Andersen said There is nothing in the SQL standard that would support the use of an '?' as anything but a parameter marker.. CREATE OPERATOR is a PostgreSQL extension. There are no provisions for user-defined operators in the SQL standard. Exactly. The standard specifies the characters to use for the predicates that it defines, and provides no mechanism for adding additional predicates; but who in the world would want to exclude all extensions to the standard? I was certainly not suggesting custom operators should be excluded. I was suggesting using something that was actually not incompatible with the SQL standards (and, even with standards aside, the expectations implementors have regarding the question mark, since it affects other tools too). And by extension if indeed the standard does require the use of ? for parameters we are in violation there because the backend protocol deals with $# placeholders and not ? We're talking about a different specification that has question marks as parameter placeholders. That's in the Java Database Connector (JDBC) specification. (It is apparently also specified in other documents, although I'm not familiar enough with those to comment.) Note that it would create all sorts of pain if both the SQL statements and a connector issuing them used the same convention for substituting parameters; it is a *good* thing that plpgsql and SQL function definitions use a different convention than JDBC! Actually, we were not just talking about JDBC. I don't know the specifications in details, but the SQL:201x (preliminary) documents linked from https://wiki.postgresql.org/wiki/Developer_FAQ#Where_can_I_get_a_copy_of_the_SQL_standards.3F seem to have some information. The Foundation document (Section 4.25 Dynamic SQL concepts) says that dynamic parameters are represented by a question mark. In addition, the BNF grammar available at http://www.savage.net.au/SQL/sql-2003-2.bnf.html#dynamic%20parameter%20specification also says: dynamic parameter specification::= question mark I'm not familiar enough with these documents to know whether I'm missing some context, but it would seem that the question mark is a reserved character, beyond the scope of JDBC (at the very least, it seems incompatible with Dynamic SQL and its implementation in ECPG). Best wishes, Bruno.
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
Dave Cramer dave.cramer(at)credativ(dot)ca http://www.credativ.ca On 19 May 2015 at 13:15, Mike Blackwell mike.blackw...@rrd.com wrote: A Google search suggests Oracle 9.x supports a unary '?' operator (fuzzy match), so the use of '?' in an operator name is not without precedent. Interesting argument. There is considerable precedent where we take the position that just because xyz supports it we don't. Dave Cramer dave.cramer(at)credativ(dot)ca http://www.credativ.ca
Re: [HACKERS] a few thoughts on the schedule
On 2015-05-19 10:25:49 -0400, Robert Haas wrote: On Mon, May 18, 2015 at 11:52 PM, Andres Freund and...@anarazel.de wrote: I personally think the late close of the 9.4 cycle has alone thrings far enough off track that we can't fairly evaluate a 5 CF schedule. Oh, I agree with that. Ah, ok. The vary earliest time frame that would make sense to me is to branch July 1st and start a CF on July 15th. I'm wondering why the CF has to start after branching? Or is that just two independent dates? The first week or so of the first CF won't have much stuff ready for commit. Personally, given where we're at right now, I don't think an early fall release of 9.5 is going to be remotely practical. Why? To me the last few beta periods were pretty drawn out affairs, without much happening. Yes, there was the jsonb stuff in 9.4 delaying the release, but that wasn't waiting for work, but a decision. But most of the time everyone was developing their stuff for the next cycle, waiting for beta testers to come forward with bugs. Not very much of that happened. I think a shorter schedule might actually help us to both, get the open issues closed sooner, and get more actual testing. Most people seem to work with a Oh, there's time left, I can do that later attitude. I mean if there'd actually be lots of people busy testing, sure, a long beta makes sense for postgres (complex, contains critical data). But I don't think that's happening. - Do 4 CommitFests as we have for past releases. We could do July 15th, September 15th, November 15th, and January 15th; or we could do August 1st, October 1st, December 1st, and February 1st; or we could do August 15th, October 15th, December 15th, and February 15th. Probably, that last one isn't so good: starting on December 15th is going to suck. I tend to agree that Dec 15 is a bad idea. Maybe we should forget them and just have monthly 'judgefests' where some poor sod summarizes the current state and direction, and we then collaboratively discuss whether we see things going anywhere and if not, what would need to happen that they do. And have a policy that older patches should be preferred over newer ones; but at the same time cull patches continually sitting at the tail end as 'not interesting'. I think we need to start by understanding that we need the contribution experience to be good for both patch authors and also for reviewers (including reviewers who are commiters). We very much need to give new contributors a good experience of submitting patches and getting useful feedback and getting stuff committed. I think it's clear that we could do a much better job at that, and the project would benefit enormously. Agreed. I think right now to succeed in the project you need to be extraordinarily stubborn or patient. Which in turn comes with its own set of problems in the long term, besides lower participation. The set of qualities needed to succeed aren't the same that I see needed in the project. However, doing a better job means spending more time on it, and we can't just demand that senior reviewers or contributors spend more time on it. I mean, we can, I guess, but it will only breed frustration and resentment. I'm not sure what the solution is here, but if it boils down to telling people who have put a lot of effort into the project over a long period of time that they are not doing enough, I'm here to say that won't work. Agreed, that we can't just demand it. But I think without changing anything the situation will just get worse and worse, because there'll be few new senior people. I think part of that is saying no more efficiently, upfront. Which is why I really want the triage step. a) It's much better for the project to not have several junior reviewers first spend time on a patch, then have a small flamefest, and then have somebody senior reject a patch in its entirety. That's a waste of everyone's effort and frustrating. b) It's not that bad to hear a no as a new contributor soon after submission. It's something entirely different to go through a long bikeshedding, several revisions of reworking, just to be told in the end that it was a bad idea from the get go. So one problem that comes up in the context of your proposal is that it's likely to be hard to find the poor sod whose existence you hypothecate. Maybe there is someone who will do that once or twice, but I think it'll be hard to keep that position filled over the long term. I'm not sure. ISTM that a painfull couple hours every now and then are much less bad than the continuous CF we had lately. I personally also find it frustrating to go through the CF and see a good portion of things that I never can see going anywhere, but that still suck up resources. I'd actually be willing to do triage every now and then; but I don't think it should always be the same person. For one it does come with power, for another it's nice to now
Re: [HACKERS] Per row status during INSERT .. ON CONFLICT UPDATE?
On Tue, May 19, 2015 at 1:07 PM, Robins Tharakan thara...@gmail.com wrote: My use-case is to create an extra row for all UPDATEd rows (only), which is implemented in MSSQL by enveloping the MERGE with an INSERT (MERGE ... OUTPUT $action) WHERE $action = 'UPDATE'. That could make sense. You can achieve something similar with per-row triggers, perhaps. Am still to test, but looks like Thom's reply earlier could take care of my use-case, so we may need more people requesting this magic field, with a valid use-case. I'm not opposed to it, but it's not a personal priority to implement this. I don't think it's great practice to use the hidden fields like that. I can't see anything other than a dedicated expression serving this purpose, if there is ever a documented solution. -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Per row status during INSERT .. ON CONFLICT UPDATE?
On Tue, May 19, 2015 at 1:20 PM, Peter Geoghegan p...@heroku.com wrote: On Tue, May 19, 2015 at 1:07 PM, Robins Tharakan thara...@gmail.com wrote: My use-case is to create an extra row for all UPDATEd rows (only), which is implemented in MSSQL by enveloping the MERGE with an INSERT (MERGE ... OUTPUT $action) WHERE $action = 'UPDATE'. That could make sense. You can achieve something similar with per-row triggers, perhaps. BTW, be prepared to deal with the updated row (*any* row version) not being visible to your MVCC snapshot with that pattern in Postgres (at READ COMMITTED level). It probably won't matter, but it could. -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On 19 May 2015 at 16:36, Geoff Winkless pgsqlad...@geoff.dj wrote: On 19 May 2015 at 21:12, Peter Geoghegan p...@heroku.com wrote: It's trivial to modify Postgres to not require that a specific unique index be inferred, so that you can omit the inference specification for DO UPDATE just as you can for DO NOTHING. That would make it work in a similar way to MySQL; whatever actually conflict was detected would be assumed to be cause to take the alternative update path. Except that would break the deterministic behaviour, surely? Because if you only updated one row based on which constraint matched first, the row that was updated would depend on the order in which the constraints were evaluated, yes? It would depend upon the evaluation order, but that would not break determinism unless you allowed a random evaluation order. Omitting the clause for DO NOTHING yet requiring it for DO UPDATE doesn't make sense. We should allow DO UPDATE to exclude a constraint and apply a deterministic order to the constraints. 1. PK if it exists. 2. Replica Identity, when not PK, 3. UNIQUE constraints in name order, like triggers, so users can define a default evaluation order, just like they do with triggers. I was expecting that matching two constraints would end up UPDATEing two separate rows. It's not clear to me how a single INSERT could cause two or more UPDATEs. I have a hard time imagining why you'd ever not want to be explicit about what to take the alternative path on for the DO UPDATE variant. What do you have in mind? If I'm being honest, my main driver is laziness :) I don't mind specifying the constraint if I can understand why it's required, but otherwise it just seems like I need to do more typing for no reason. Especially when there's only one unique constraint on a table. 1) Ease of use - Unique constraints don't change very often. This saves time for the common case where they stay the same. It also saves time if they do change, because you avoid having to completely recode your app AND make that happen at exactly the same time you apply the change of unique constraint. 2) Compatibility with MySQL -- Simon Riggshttp://www.2ndQuadrant.com/ http://www.2ndquadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training Services
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On 19 May 2015 at 17:10, Peter Geoghegan p...@heroku.com wrote: On Tue, May 19, 2015 at 1:57 PM, Simon Riggs si...@2ndquadrant.com wrote: We should allow DO UPDATE to exclude a constraint and apply a deterministic order to the constraints. 1. PK if it exists. 2. Replica Identity, when not PK, 3. UNIQUE constraints in name order, like triggers, so users can define a default evaluation order, just like they do with triggers. That seems like something way worse than just allowing it for all constraints. I'm talking about the evaluation order; it would still match all constraints, otherwise they wouldn't be constraints. 2) Compatibility with MySQL But what you describe isn't compatible with MySQL. It's totally novel. Upthread you said It's trivial to modify Postgres to not require that a specific unique index be inferred, so that you can omit the inference specification for DO UPDATE just as you can for DO NOTHING. That would make it work in a similar way to MySQL Similar is good and useful. Full compatibility is even better. -- Simon Riggshttp://www.2ndQuadrant.com/ http://www.2ndquadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training Services
Re: [HACKERS] Bug in jsonb minus operator
On 05/18/2015 10:52 PM, Peter Geoghegan wrote: On Mon, May 18, 2015 at 7:11 AM, Andrew Dunstan and...@dunslane.net wrote: Here's an patch along those lines. It seems to do the trick, at least for your test case, and it has the merit of being very small, so small I'd like to backpatch it - accepting jbvBinary as is in pushJsonbValue seems like a bug to me. Isn't that for the benefit of raw scalar pseudo arrays? The existing comments above pushJsonbValue() acknowledge such callers. Umm, no, the raw scalar pseudo arrays are of type jbvArray, not jbvBinary. And they are pushed with WJB_BEGIN_ARRAY, not with WJB_ELEM or WJB_VALUE, as the comment notes. See this fragment of code from JsonbValueToJsonb: scalarArray.type = jbvArray; scalarArray.val.array.rawScalar = true; scalarArray.val.array.nElems = 1; pushJsonbValue(pstate, WJB_BEGIN_ARRAY, scalarArray); I tested this by removing the assert test for jbvBinary in the WJB_ELEM and WJB_VALUE switch beranches and then running the regression suite. No assertion failure was triggered. While that's not guaranteed to be a perfect test, it doesn't seem like a bad one. Can you pose a counter example where this will break? Why are you passing the skipNested variable to JsonbIteratorNext() within jsonb_delete()? I'm not seeing a need for that. If you don't the logic gets more complex, as you need to keep track of what level of the object you are at. The virtue of this change is that it will simplify a lot of such processing by removing the unnecessary restriction on passing jbvBinary values to pushJsonbValue(). If you have a better fix for the bug you complained about I'll be happy to take a look. cheers andrew -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On Tue, May 19, 2015 at 12:57 PM, Geoff Winkless pgsqlad...@geoff.dj wrote: Well http://www.postgresql.org/docs/devel/static/sql-insert.html explains that a conflict_target clause is required but doesn't explain why. Yes, for ON CONFLICT DO UPDATE, it is mandatory. It _does_ make clear that multiple UPDATEs to the same row are not allowed, but that in itself doesn't automatically restrict the use of multiple constraint targets; I could easily INSERT a set of values that would trigger that failure with just one constraint target. True. http://www.postgresql.org/docs/devel/static/sql-insert.html talks about how MySQL's ON DUPLICATE can only act against the first matching row where multiple constraints match against multiple rows. I suppose if that were the case here (ie the first excluding row would stop other rows firing against the UPDATE) would break the deterministic feature, but it's not clear if that's true or not. I don't see why multiple target rows couldn't be updated based on multiple constraints, that would not in-and-of-itself break determinism. If I'm missing the obvious, accept my apologies. It's trivial to modify Postgres to not require that a specific unique index be inferred, so that you can omit the inference specification for DO UPDATE just as you can for DO NOTHING. That would make it work in a similar way to MySQL; whatever actually conflict was detected would be assumed to be cause to take the alternative update path. The only reason I can see for wanting to do this is where you're running a migration or something, and two unique indexes are equivalent anyway. Like maybe you have a partial index and a non-partial index, and you're just about to drop one of them. But the inference specification will do the right thing here anyway -- multiple unique indexes can be inferred for edge cases like this. I have a hard time imagining why you'd ever not want to be explicit about what to take the alternative path on for the DO UPDATE variant. Unless perhaps you have a different UPDATE targetlist and so on corresponding to that case, which is currently not possible -- but then what if multiple constraints have would-be violations at the same time? It gets awfully messy very quickly. What do you have in mind? -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
On Tue, May 19, 2015 at 3:00 PM, Simon Riggs si...@2ndquadrant.com wrote: As long as the cookie is randomly generated for each use, then I don't see a practical problem with that approach. If the client sets the cookie via an SQL command, that command would be written to the log, and displayed in pg_stat_activity. A malicious user might be able to get it from one of those places. A malicious user might also be able to just guess it. I don't really want to create a situation where any weakess in pgpool's random number generation becomes a privilege-escalation attack. A protocol extension avoids all of that trouble, and can be target for 9.6 just like any other approach we might come up with. I actually suspect the protocol extension will be FAR easier to fully secure, and thus less work, not more. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
[HACKERS] Rewriting backup.sgml (patch attached)
Hello, Alright, per previous discussions I went through the backup.sgml page. I have gone thoroughly through: sql dump pg_dump pg_restore handling large databases I removed file based backups I didn't really touch the red headed step child that is pg_dumpall (although a word smithed it a little). I tried to remove some of the conversationalist tone. This is technical documentation not a story. I also removed as many extra words as reasonable and added specific examples. Now, I know some of you will say things like... why did you add -C to the pg_dump line. The reason is simple, without it as new person is going to get an error on restore. Our documentation should be explicit not implicit. I do not want to progress any farther on this until I get some feedback. Sincerely, JD -- Command Prompt, Inc. - http://www.commandprompt.com/ 503-667-4564 PostgreSQL Centered full stack support, consulting and development. Announcing I'm offended is basically telling the world you can't control your own emotions, so everyone else should do it for you. diff --git a/doc/src/sgml/backup.sgml b/doc/src/sgml/backup.sgml index def43a2..cdc288b 100644 --- a/doc/src/sgml/backup.sgml +++ b/doc/src/sgml/backup.sgml @@ -6,19 +6,17 @@ indexterm zone=backupprimarybackup// para - As with everything that contains valuable data, productnamePostgreSQL/ - databases should be backed up regularly. While the procedure is - essentially simple, it is important to have a clear understanding of + productnamePostgreSQL/ databases should be backed up regularly. + The procedure is simple but it is important to have a clear understanding of the underlying techniques and assumptions. /para para - There are three fundamentally different approaches to backing up + There are four different approaches to backing up productnamePostgreSQL/ data: itemizedlist listitemparaacronymSQL/ dump/para/listitem - listitemparaFile system level backup/para/listitem - listitemparaContinuous archiving/para/listitem + listitemparaacronymPITR/, Point in Time Recovery/para/listitem /itemizedlist Each has its own strengths and weaknesses; each is discussed in turn in the following sections. @@ -28,141 +26,176 @@ titleacronymSQL/ Dump/title para - The idea behind this dump method is to generate a file with SQL - commands that, when fed back to the server, will recreate the - database in the same state as it was at the time of the dump. - productnamePostgreSQL/ provides the utility program - xref linkend=app-pgdump for this purpose. The basic usage of this - command is: + productnamePostgreSQL/ provides the program xref linkend=app-pgdump for + generating a backup file with SQL commands that, when fed back to the server, + will recreate the database in the same state as it was at the time of the dump. + The basic usage of xref linkend=app-pgdump is: synopsis -pg_dump replaceable class=parameterdbname/replaceable gt; replaceable class=parameteroutfile/replaceable +pg_dump replaceable class=parameter-C/replaceable replaceable +class=parameter-F/replaceable optionp/option replaceable +class=parameter-f/replaceable optionoutfile/option optiondbname/option /synopsis - As you see, applicationpg_dump/ writes its result to the - standard output. We will see below how this can be useful. - While the above command creates a text file, applicationpg_dump/ - can create files in other formats that allow for parallism and more - fine-grained control of object restoration. + + The use of option-C/option ensures that the dump file will + contain the requisite xref linkend=sql-createdatabase command within the dump file. The use of + replaceable class=parameter-F/replaceableoptionp/option ensures that you are using the plain + text format and the use of replaceable class=parameter-f/replaceable allows you + to specify the name of the file the dump will be written to. It is also possible + for applicationpg_dump/ to create files in other formats that allow for parallelism + and fine-grained control of object backup or restoration. For more details on all options + available to applicationpg_dump/ please refer to the xref linkend=app-pgdump reference page. /para para - applicationpg_dump/ is a regular productnamePostgreSQL/ - client application (albeit a particularly clever one). This means - that you can perform this backup procedure from any remote host that has - access to the database. But remember that applicationpg_dump/ - does not operate with special permissions. In particular, it must - have read access to all tables that you want to back up, so in order - to back up the entire database you almost always have to run it as a - database superuser. (If you do not have sufficient privileges to back up - the entire database, you can still back up portions of the database to which - you do have access using options such as -
Re: [HACKERS] Minor ON CONFLICT related fixes
On 2015-05-18 19:09:27 -0700, Peter Geoghegan wrote: On Mon, May 18, 2015 at 2:09 PM, Peter Geoghegan p...@heroku.com wrote: You pointed out that the reason for this trivial bug on Jabber, but here's the obvious fix, including an EXPLAIN regression test. Also, I attach a patch adding ruleutils.c deparsing support for INSERT statement target aliases. This was previously overlooked. Pushed. -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Per row status during INSERT .. ON CONFLICT UPDATE?
On 19 May 2015 at 23:24, Peter Geoghegan p...@heroku.com wrote: That's certainly something we talked about. It could probably be done with some kind of magical expression. I have to wonder how many of the people that are sure that they need this really do, though. Is it really natural to care about this distinction with idiomatic usage? Thanks everyone for responding promptly. Not sure if I can be authoritative for many, but for me, the need emanates from having to move an ETL off MSSQL Server, which supports OUTPUT $action (similar to RETURNING * in Postgres) where $action is the per-row status (INSERT / UPDATE). My use-case is to create an extra row for all UPDATEd rows (only), which is implemented in MSSQL by enveloping the MERGE with an INSERT (MERGE ... OUTPUT $action) WHERE $action = 'UPDATE'. Am still to test, but looks like Thom's reply earlier could take care of my use-case, so we may need more people requesting this magic field, with a valid use-case. -- Robins Tharakan
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
On 05/19/2015 09:00 PM, Simon Riggs wrote: [snip] I think the idea of having SET SESSION AUTH pass a cookie, and only let RESET SESSION AUTH work when the same cookie is supplied, is pretty reasonable. As long as the cookie is randomly generated for each use, then I don't see a practical problem with that approach. Protocol level solution means we have to wait 1.5 years before anybody can begin using that. I'm also dubious that a small hole in the protocol arrangements could slam that door shut because we couldn't easily backpatch. Having an in-core pooler would be just wonderful because then we could more easily trust it and we wouldn't need to worry. Ufff Please don't do that. Postgres is just a database. And a very good one at that. Let us keep it that way and not try to re-implement everything within it --- We're not the big red company after all :) There are places where a pooler is badly needed and others where it is just overkill and counterproductive. Plus, scalability models / usage patterns are not nearly the same (nor even compatible sometimes!) between databases and poolers. There exist perfectly good solutions already (and they can certainly be improved), such as PgBouncer (or even PgPool-II) or others can be adopted. Just my .02€ / J.L.
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On Tue, May 19, 2015 at 1:36 PM, Kevin Grittner kgri...@ymail.com wrote: Gavin Flower gavinflo...@archidevsys.co.nz wrote: I prefer the $1 approach, others can't use that, and there are situations where I could not either. So, how about defaulting to the '?' approach, but have a method to explicitly set the mode - to switch to using '$'? Are you suggesting that we implement something other than what is described in these documents for prepared statement parameters?: http://docs.oracle.com/javase/7/docs/api/java/sql/PreparedStatement.html http://download.oracle.com/otn-pub/jcp/jdbc-4_1-mrel-spec/jdbc4.1-fr-spec.pdf If so, I strongly oppose that. If we are not going to deprecate use of the question mark character for operators, we need some nonstandard hack to our JDBC implementation, but an alternative syntax for specifying PreparedStatement and CallableStatement parameters seems entirely the wrong way to go. I'll repeat my earlier comment that having a mode that allows for libpq syntax while still conforming to the JDBC class API would have value for those users willing to admit their application and code is not portable (and if they are using these operators it is not) and would rather conform as closely to native PostgreSQL language mechanics as possible. That said I would not argue that the current official driver needs to be so modified. The issue here is what to do about the difficulties in using JDBC prepared statements in combination with the PostgreSQL extension of operator names containing question marks. Using a double question mark is not horrible as a solution. It may not be what we would have arrived at had the discussion taken place on the pgsql-jdbc list rather than underneath a github pull request, but we can only move forward from where we are. Out of curiosity, how long has the ?? solution been implemented in a driver jar file available as a public download? Less than 6 months...discussion started a few months prior to that. What are the guidelines for what discussion belongs on the pgsql-jdbc list and what discussion belongs on github? Is someone interested in participating in the discussions leading to decisions about our JDBC connector expected to follow both? As things stand now - it seems that way. There are no guidelines that I can tell but I'd likely consider pgsql-jdbc the equivalent of -general and GitHub looks like -hackers. Neither is particularly high volume. David J.
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On Tue, May 19, 2015 at 1:36 PM, Geoff Winkless pgsqlad...@geoff.dj wrote: On 19 May 2015 at 21:12, Peter Geoghegan p...@heroku.com wrote: It's trivial to modify Postgres to not require that a specific unique index be inferred, so that you can omit the inference specification for DO UPDATE just as you can for DO NOTHING. That would make it work in a similar way to MySQL; whatever actually conflict was detected would be assumed to be cause to take the alternative update path. Except that would break the deterministic behaviour, surely? Because if you only updated one row based on which constraint matched first, the row that was updated would depend on the order in which the constraints were evaluated, yes? I was expecting that matching two constraints would end up UPDATEing two separate rows. Well, it would be deterministic to the extent that the indexes would be evaluated in OID order. But yes, the first would-be duplicate violation would make the update path be taken once and only once for the row proposed for insertion -- at that point, you've given up on insertion (unless there is a row locking conflict). Just like MySQL, I believe. How can you find a would-be violation without inserting? How can you insert without also violating the other thing? It's far messier than it first appears. I have a hard time imagining why you'd ever not want to be explicit about what to take the alternative path on for the DO UPDATE variant. What do you have in mind? If I'm being honest, my main driver is laziness :) I don't mind specifying the constraint if I can understand why it's required, but otherwise it just seems like I need to do more typing for no reason. Especially when there's only one unique constraint on a table. Well, I don't have zero sympathy for that, but I'm pretty sure that that's what other people wanted. If I'm being honest, I don't actually remember how true that was. -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On Tue, May 19, 2015 at 2:34 PM, Bruno Harbulot br...@distributedmatter.net wrote: While I can imagine a Java PostgreSQL driver that would use the libpq syntax, I can't see it being able to have any useful sort of half-compatibility with JDBC, whether it mimics its interfaces or not. I'm not sure it would be very useful at all, considering how much the existing tooling the the Java world relies on JDBC. I won't claim to have studied this in great detail but there is a lot more to the JDBC spec beyond the semantics of PreparedStatement.parse(String). No need to throw out the baby with the bath water and reinvent ResultSet, Connection and various other interfaces that are perfectly usable before and after a suitable query has been fully parsed. When I say setInteger(1, new Integer(1000)) I don't care whether I had to write SELECT ? AS int_val OR SELECT $1 AS int_val; though the later has the nice property of providing corresponding numbers so that I would write something like SELECT $1 AS int_val, $1 AS int_val_2 and not be forced to write setInteger(2, new Integer(1000)) to pass in a value to the second - but identical - parameter. Maybe it violates the semantics defined by the API - which I could be making too lightly of - but having the same mechanics involved to solve the same problem - with only minor semantic nuances to remember seems within the realm of reasonable. David J.
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
Gavin Flower gavinflo...@archidevsys.co.nz wrote: I prefer the $1 approach, others can't use that, and there are situations where I could not either. So, how about defaulting to the '?' approach, but have a method to explicitly set the mode - to switch to using '$'? Are you suggesting that we implement something other than what is described in these documents for prepared statement parameters?: http://docs.oracle.com/javase/7/docs/api/java/sql/PreparedStatement.html http://download.oracle.com/otn-pub/jcp/jdbc-4_1-mrel-spec/jdbc4.1-fr-spec.pdf If so, I strongly oppose that. If we are not going to deprecate use of the question mark character for operators, we need some nonstandard hack to our JDBC implementation, but an alternative syntax for specifying PreparedStatement and CallableStatement parameters seems entirely the wrong way to go. The issue here is what to do about the difficulties in using JDBC prepared statements in combination with the PostgreSQL extension of operator names containing question marks. Using a double question mark is not horrible as a solution. It may not be what we would have arrived at had the discussion taken place on the pgsql-jdbc list rather than underneath a github pull request, but we can only move forward from where we are. Out of curiosity, how long has the ?? solution been implemented in a driver jar file available as a public download? What are the guidelines for what discussion belongs on the pgsql-jdbc list and what discussion belongs on github? Is someone interested in participating in the discussions leading to decisions about our JDBC connector expected to follow both? -- Kevin Grittner EDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On 19 May 2015 at 21:12, Peter Geoghegan p...@heroku.com wrote: It's trivial to modify Postgres to not require that a specific unique index be inferred, so that you can omit the inference specification for DO UPDATE just as you can for DO NOTHING. That would make it work in a similar way to MySQL; whatever actually conflict was detected would be assumed to be cause to take the alternative update path. Except that would break the deterministic behaviour, surely? Because if you only updated one row based on which constraint matched first, the row that was updated would depend on the order in which the constraints were evaluated, yes? I was expecting that matching two constraints would end up UPDATEing two separate rows. I have a hard time imagining why you'd ever not want to be explicit about what to take the alternative path on for the DO UPDATE variant. What do you have in mind? If I'm being honest, my main driver is laziness :) I don't mind specifying the constraint if I can understand why it's required, but otherwise it just seems like I need to do more typing for no reason. Especially when there's only one unique constraint on a table. Geoff
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
On Tue, May 19, 2015 at 2:46 PM, Andres Freund and...@anarazel.de wrote: On 2015-05-19 14:41:06 -0400, Robert Haas wrote: On Tue, May 19, 2015 at 12:29 PM, Andres Freund and...@anarazel.de wrote: On 2015-05-19 10:53:10 -0400, Robert Haas wrote: That seems like a kludge to me. If the cookie leaks out somhow, which it will, then it'll be insecure. I think the way to do this is with a protocol extension that poolers can enable on request. Then they can just refuse to forward any reset authorization packets they get from their client. There's no backward-compatibility break because the pooler can know, from the server version, whether the server is new enough to support the new protocol messages. That sounds like a worse approach to me. Don't you just need to hide the session authorization bit in a function serverside to circumvent that? I'm apparently confused. There's nothing you can do to maintain security against someone who can load C code into the server. I must be misunderstanding you. It very well might be me that's confused. But what's stopping a user from doing a RESET SESSION AUTHORIZATION; in a DO block or something? I guess you are intending that a RESET SESSION AUTHORIZATION is only allowed on a protocol level when the protocol extension is in use? Yes, something like that. I'm not sure if we'd want to reuse the existing SESSION AUTHORIZATION concept or create something new, but either way the idea would be that the pooler would send a PoolerSetAuthorization message which could only be undone by another such message. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
-BEGIN PGP SIGNED MESSAGE- Hash: RIPEMD160 Jan de Visser wrote: Well, one could argue that it *is* their problem, as they should be using the standard Postgres way for placeholders, which is $1, $2, $3... Shirley you are joking: Many products use JDBC as an abstraction layer facilitating (mostly) seamless switching between databases. I know the product I worked on did. Are you advocating that every single statement should use SELECT * FROM foo WHERE bar = $1 on pg and SELECT * FROM foo WHERE bar = ? on every other database? I'm not joking, and don't call me Shirley. If you are running into situations where you have question mark operators in your queries, you have already lost the query abstraction battle. There will be no seamless switching if you are using jsonb, hstore, ltree, etc. My statement was more about pointing out that Postgres already offers a complete placeholder system, which drivers are free to implement if they want. A database is only as valuable as the the part of the outside world it can interact with. Large parts of the data-consuming world are developed in java using JDBC. If your opinion is that JDBC developers should adapt themselves to pg then you instantaneously diminish the value of pg. Well, they will have to adapt to one way or another: using ?? or \? is doing so, and the other solution (Postgres adapting itself to the driver by deprecating the ? operator) is not realistically likely to happen. - -- Greg Sabino Mullane g...@turnstep.com End Point Corporation http://www.endpoint.com/ PGP Key: 0x14964AC8 201505191718 http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8 -BEGIN PGP SIGNATURE- iEYEAREDAAYFAlVbq4AACgkQvJuQZxSWSsgrXgCaA6MTvbDeg2aMf+/HFnxutrqH P1sAoLZB1w5+UXHMxXqW/Ex0q7GwoFds =IOpS -END PGP SIGNATURE- -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On 20/05/15 07:37, Jan de Visser wrote: On May 19, 2015 07:04:56 PM Greg Sabino Mullane wrote: Bruno Harbulot asked for a devil's advocate by saying: My main point was that this is not specific to JDBC. Considering that even PostgreSQL's own ECPG is affected, the issue goes probably deeper than it seems. I'm just not convinced that passing the problem onto connectors, libraries and ultimately application developers is the right thing to do here. Well, one could argue that it *is* their problem, as they should be using the standard Postgres way for placeholders, which is $1, $2, $3... Shirley you are joking: Many products use JDBC as an abstraction layer facilitating (mostly) seamless switching between databases. I know the product I worked on did. Are you advocating that every single statement should use SELECT * FROM foo WHERE bar = $1 on pg and SELECT * FROM foo WHERE bar = ? on every other database? A database is only as valuable as the the part of the outside world it can interact with. Large parts of the data-consuming world are developed in java using JDBC. If your opinion is that JDBC developers should adapt themselves to pg then you instantaneously diminish the value of pg. jan I prefer the $1 approach, others can't use that, and there are situations where I could not either. So, how about defaulting to the '?' approach, but have a method to explicitly set the mode - to switch to using '$'? Cheers, Gavin -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On 19 May 2015 at 20:11, Simon Riggs si...@2ndquadrant.com wrote: I'm sure we'll be asked these questions many times. Can you comment on whether the docs are sufficiently detailed to explain this answer? Well http://www.postgresql.org/docs/devel/static/sql-insert.html explains that a conflict_target clause is required but doesn't explain why. It _does_ make clear that multiple UPDATEs to the same row are not allowed, but that in itself doesn't automatically restrict the use of multiple constraint targets; I could easily INSERT a set of values that would trigger that failure with just one constraint target. http://www.postgresql.org/docs/devel/static/sql-insert.html talks about how MySQL's ON DUPLICATE can only act against the first matching row where multiple constraints match against multiple rows. I suppose if that were the case here (ie the first excluding row would stop other rows firing against the UPDATE) would break the deterministic feature, but it's not clear if that's true or not. I don't see why multiple target rows couldn't be updated based on multiple constraints, that would not in-and-of-itself break determinism. If I'm missing the obvious, accept my apologies. Geoff
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On 19 May 2015 at 16:36, Kevin Grittner kgri...@ymail.com wrote: Gavin Flower gavinflo...@archidevsys.co.nz wrote: I prefer the $1 approach, others can't use that, and there are situations where I could not either. So, how about defaulting to the '?' approach, but have a method to explicitly set the mode - to switch to using '$'? Are you suggesting that we implement something other than what is described in these documents for prepared statement parameters?: http://docs.oracle.com/javase/7/docs/api/java/sql/PreparedStatement.html http://download.oracle.com/otn-pub/jcp/jdbc-4_1-mrel-spec/jdbc4.1-fr-spec.pdf If so, I strongly oppose that. If we are not going to deprecate use of the question mark character for operators, we need some nonstandard hack to our JDBC implementation, but an alternative syntax for specifying PreparedStatement and CallableStatement parameters seems entirely the wrong way to go. The issue here is what to do about the difficulties in using JDBC prepared statements in combination with the PostgreSQL extension of operator names containing question marks. Using a double question mark is not horrible as a solution. Actually the issue is what to do about a number of connectors which use a fairly standard '?' as a placeholder. Notably absent from the discussion is ODBC upon which JDBC was modelled and probably predates any use of ? as an operator It may not be what we would have arrived at had the discussion taken place on the pgsql-jdbc list rather than underneath a github pull request, but we can only move forward from where we are. possibly, however all of the current JDBC maintainers opined and reached an agreement on this. Out of curiosity, how long has the ?? solution been implemented in a driver jar file available as a public download? At least since February of this year What are the guidelines for what discussion belongs on the pgsql-jdbc list and what discussion belongs on github? Is someone interested in participating in the discussions leading to decisions about our JDBC connector expected to follow both? Currently pull requests are the easiest to deal with so most discussion is on github. I guess updating the JDBC web page would be in order. Dave Cramer dave.cramer(at)credativ(dot)ca http://www.credativ.ca
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
On 19 May 2015 at 16:49, Robert Haas robertmh...@gmail.com wrote: On Tue, May 19, 2015 at 3:00 PM, Simon Riggs si...@2ndquadrant.com wrote: As long as the cookie is randomly generated for each use, then I don't see a practical problem with that approach. If the client sets the cookie via an SQL command, that command would be written to the log, and displayed in pg_stat_activity. A malicious user might be able to get it from one of those places. A malicious user might also be able to just guess it. I don't really want to create a situation where any weakess in pgpool's random number generation becomes a privilege-escalation attack. A protocol extension avoids all of that trouble, and can be target for 9.6 just like any other approach we might come up with. I actually suspect the protocol extension will be FAR easier to fully secure, and thus less work, not more. That's a reasonable argument. So +1 to protocol from me. To satisfy Tom, I think this would need to have two modes: one where the session can never be reset, for ultra security, and one where the session can be reset, which allows security and speed of pooling. -- Simon Riggshttp://www.2ndQuadrant.com/ http://www.2ndquadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training Services
Re: [HACKERS] a few thoughts on the schedule
On Tue, May 19, 2015 at 1:35 PM, Andres Freund and...@anarazel.de wrote: The vary earliest time frame that would make sense to me is to branch July 1st and start a CF on July 15th. I'm wondering why the CF has to start after branching? Or is that just two independent dates? The first week or so of the first CF won't have much stuff ready for commit. Well, if there is something ready to commit, I'd like to be able to commit it. Personally, given where we're at right now, I don't think an early fall release of 9.5 is going to be remotely practical. Why? To me the last few beta periods were pretty drawn out affairs, without much happening. Yes, there was the jsonb stuff in 9.4 delaying the release, but that wasn't waiting for work, but a decision. But most of the time everyone was developing their stuff for the next cycle, waiting for beta testers to come forward with bugs. Not very much of that happened. I think a shorter schedule might actually help us to both, get the open issues closed sooner, and get more actual testing. Most people seem to work with a Oh, there's time left, I can do that later attitude. There's something to that theory. I'm just worried all of those last minute commits are hiding a bunch of bugs. I think part of that is saying no more efficiently, upfront. Which is why I really want the triage step. a) It's much better for the project to not have several junior reviewers first spend time on a patch, then have a small flamefest, and then have somebody senior reject a patch in its entirety. That's a waste of everyone's effort and frustrating. b) It's not that bad to hear a no as a new contributor soon after submission. It's something entirely different to go through a long bikeshedding, several revisions of reworking, just to be told in the end that it was a bad idea from the get go. I agree this would help. Figuring out how to do it in a reasonable way would help a lot. If we could get say 4 committers to go through at the start of each CommitFest and each comment very briefly on 25% of the patches each (yes, no, or maybe, and a bit of justification), I bet that would streamline things considerably. If we could get each committer to go through 50% of the patches and do this, then each patch would get a quick opinion from two committers right at the outset. That would be even nicer. Unless talented reviewers can get such job offers, we are going to continue to have trouble making ends meet. Hasn't every talented reviewer gotten job offers shortly afterwards in the last few years? The ones that accept don't necessarily work that much in the community, but several seem to. And I think in the case of several people the reason they don't, is less the company, but that it's emotionally draining. Agreed, lots of people get job offers. But somehow we're still short of reviewers, so something's not working the way it needs to. Maybe that's for the reason you postulate, or maybe it's for the reason I postulate, or maybe there is some other reason. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On Tue, May 19, 2015 at 1:57 PM, Simon Riggs si...@2ndquadrant.com wrote: We should allow DO UPDATE to exclude a constraint and apply a deterministic order to the constraints. 1. PK if it exists. 2. Replica Identity, when not PK, 3. UNIQUE constraints in name order, like triggers, so users can define a default evaluation order, just like they do with triggers. That seems like something way worse than just allowing it for all constraints. I have a hard time imagining why you'd ever not want to be explicit about what to take the alternative path on for the DO UPDATE variant. What do you have in mind? If I'm being honest, my main driver is laziness :) I don't mind specifying the constraint if I can understand why it's required, but otherwise it just seems like I need to do more typing for no reason. Especially when there's only one unique constraint on a table. 1) Ease of use - Unique constraints don't change very often. This saves time for the common case where they stay the same. It also saves time if they do change, because you avoid having to completely recode your app AND make that happen at exactly the same time you apply the change of unique constraint. I don't see how it's possible to change unique constraints in a way that breaks the inference specification without that actually being desirable -- naming the constraint by name is an escape hatch that is generally discouraged. That's the whole point of inference. I put an awful lot of work into making unique index inference as forgiving as possible. For example, it doesn't care what order attributes appear in, or if they appear redundantly, or if an ON CONFLICT unique index predicate is more selective than any available index that is otherwise satisfied (there is a call to predicate_implied_by()). 2) Compatibility with MySQL But what you describe isn't compatible with MySQL. It's totally novel. -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On Tue, May 19, 2015 at 9:50 PM, David G. Johnston
david.g.johns...@gmail.com wrote:
On Tue, May 19, 2015 at 1:36 PM, Kevin Grittner kgri...@ymail.com wrote:
Gavin Flower gavinflo...@archidevsys.co.nz wrote:
I prefer the $1 approach, others can't use that, and there are
situations where I could not either.
So, how about defaulting to the '?' approach, but have a method
to explicitly set the mode - to switch to using '$'?
Are you suggesting that we implement something other than what is
described in these documents for prepared statement parameters?:
http://docs.oracle.com/javase/7/docs/api/java/sql/PreparedStatement.html
http://download.oracle.com/otn-pub/jcp/jdbc-4_1-mrel-spec/jdbc4.1-fr-spec.pdf
If so, I strongly oppose that. If we are not going to deprecate
use of the question mark character for operators, we need some
nonstandard hack to our JDBC implementation, but an alternative
syntax for specifying PreparedStatement and CallableStatement
parameters seems entirely the wrong way to go.
I'll repeat my earlier comment that having a mode that allows for libpq
syntax while still conforming to the JDBC class API would have value for
those users willing to admit their application and code is not portable
(and if they are using these operators it is not) and would rather conform
as closely to native PostgreSQL language mechanics as possible.
I don't think that approach is workable at all. JDBC isn't limited to a
number of classes and their methods, the documentation that surrounds it
obviously has an impact on how it was implemented internally and what users
should and shouldn't be allowed to expect when using these classes. While
there are tools that convert various parameter styles to ? (e.g. Groovy SQL
or Hibernate's named parameter) and a layer of conversion from $1 to ?
could exist, the bottleneck here will still be the JDBC layer itself, since
it's what sends the query to the database.
Users of question mark operators are already admitting their application
and code isn't portable (since they are specific to PostgreSQL and its
extensions). The problem has more to do with how the other tools around
handle these customisations. For example, it can be useful to have a model
based on Hibernate in Java and be able to use ? operators for specific
features. (Other tools like SQLAlchemy in Python also allow you to have
customisations specific to the RDMBS platform, while being able to use the
core features in a more platform-neutral way.)
It turns out that you can indeed use ? in JSONB with a custom Hibernate
query, you just need to double-escape it as follows: ? becomes ?? and has
to be escaped as \?\?, but \ has to be escaped itself...
SQLQuery query = session
.createSQLQuery(SELECT
CAST((CAST('{\key1\:123,\key2\:\Hello\}' AS jsonb) \\?\\? CAST(? AS
text)) AS BOOLEAN));
query.setString(0, key1);
Again, this may have to do with the fact that these tools may have a
legitimate expectation that ? should be reserved for parameters, partly
because it seems to be very common in practice, but more importantly if the
SQL specification itself says it's what ? is for.
While I can imagine a Java PostgreSQL driver that would use the libpq
syntax, I can't see it being able to have any useful sort of
half-compatibility with JDBC, whether it mimics its interfaces or not. I'm
not sure it would be very useful at all, considering how much the existing
tooling the the Java world relies on JDBC.
This problem is also broader than JDBC: on top of the languages and
libraries already mentioned, it may affect ODBC, as Dave Cramer has just
said (I haven't tried).
Best wishes,
Bruno.
[HACKERS] Change pg_cancel_*() to ignore current backend
I find it annoying to have to specifically exclude pg_backend_pid() from pg_stat_activity if I'm trying to kill a bunch of backends at once, and I can't think of any reason why you'd ever want to call a pg_cancel_* function with your own PID. Any objections to modifying those functions so they do nothing when handed the PID of the current backend? -- Jim Nasby, Data Architect, Blue Treble Consulting Data in Trouble? Get it in Treble! http://BlueTreble.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
[HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
I finally got around to running some UPSERT tests on the development build, which is very exciting for me :) I'm not sure if I missed the point with this (probably...): I'm unclear on the reason why DO UPDATE requires explicitly specifying the constraint while DO NOTHING does not. If it's a feature of the locking implementation (or something) that for DO UPDATE only one index can be used, then so be it. However if it would be possible to allow any conflict to run the UPDATE clause (in the same way as any conflict triggers DO NOTHING in the alternate form) I would personally find that very pleasant. You could even then arbitrate on conflicts in the UPDATE clause, if you had to, using (say) INSERT INTO mytable ... ON CONFLICT DO UPDATE SET col1=CASE WHEN mytable.uniquefield1=excluded.uniquefield1 THEN targettedvalue1 ELSE mytable.col1 END, col2=CASE WHEN mytable.uniquefield2=excluded.uniquefield2 THEN targettedvalue2 ELSE mytable.col2 END; Not exactly pretty but workable. I just find it slightly upsetting that for (what I would expect is) the majority use case (when the INSERT would only ever trigger one unique constraint) one must still define the unique fields. In the event that the INSERT triggers a constraint that the UPDATE fails to resolve, it will still fail in exactly the same way that running the ON CONFLICT on a specific constraint would fail, so it's not like you gain any extra value from specifying the constraint, is it? As I said, I probably missed the point. Geoff
Re: [HACKERS] Wrong Assert in PageIndexMultiDelete?
On 05/19/2015 07:15 PM, Tom Lane wrote: Anastasia Lubennikova lubennikov...@gmail.com writes: I am trying to create new index access method. And I found strange Assert in PageIndexMultiDelete http://doxygen.postgresql.org/bufpage_8c_source.html#l00791 function. Assert http://doxygen.postgresql.org/c_8h.html#a706ac5b1a53bd04067f81924b92cb9f6(nitems MaxIndexTuplesPerPage http://doxygen.postgresql.org/itup_8h.html#adb7c94e95ce112eb47d71f7797604ddb ); Is '' sign is correct? I thougt it should be '='. Is it a bug or just my misunderstanding? Hm, I think it's a bug. It's probably not very significant because MaxIndexTuplesPerPage is an overestimate (it doesn't account for index special space), but it's wrong AFAICS. That Assert hasn't been there very long, either --- seems to have been added in 877b0887. Heikki, did you have some specific reason for writing it like that? No, it looks like a bug to me as well. Will fix... - Heikki -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
[HACKERS] errmsg() clobbers errno
Hi All While debugging an extension I discovered that the errmsg() function zeros out errno. This is annoying because if the process of assembling a meaningful error message happens to call errmsg() before calling strerror() we lose the strerror information. This is exactly the time when we want to preserve any available error state. I am attaching a patch to preserve errno across errmsg() calls. Does this seem like a good idea? Best, John errmsg-errno-v1.patch Description: Binary data -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Wrong Assert in PageIndexMultiDelete?
Anastasia Lubennikova lubennikov...@gmail.com writes: I am trying to create new index access method. And I found strange Assert in PageIndexMultiDelete http://doxygen.postgresql.org/bufpage_8c_source.html#l00791 function. Assert http://doxygen.postgresql.org/c_8h.html#a706ac5b1a53bd04067f81924b92cb9f6(nitems MaxIndexTuplesPerPage http://doxygen.postgresql.org/itup_8h.html#adb7c94e95ce112eb47d71f7797604ddb ); Is '' sign is correct? I thougt it should be '='. Is it a bug or just my misunderstanding? Hm, I think it's a bug. It's probably not very significant because MaxIndexTuplesPerPage is an overestimate (it doesn't account for index special space), but it's wrong AFAICS. That Assert hasn't been there very long, either --- seems to have been added in 877b0887. Heikki, did you have some specific reason for writing it like that? regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Per row status during INSERT .. ON CONFLICT UPDATE?
On 2015-05-19 17:53:09 +0530, Robins Tharakan wrote: Is there a way to know which rows were INSERTed and UPDATEd when doing a INSERT ... ON CONFLICT UPDATE? Probably via pseudo column indicating INSERT / UPDATE ? No, not really. The RETURNING clause just allows us to return columns, but am unable to find a way to know 'what' happened to a given row. There previously has been discussion about extending RETURNING to allow to return the before/after row. But to me that's a mostly independent feature to ON CONFLICT. Greetings, Andres Freund -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
See for example http://docs.oracle.com/cd/B19306_01/text.102/b14218/cqoper.htm#i997330, Table 3-1, third row, showing the precedence of '?'. Further down the page, under Fuzzy see Backward Compatibility Syntax. __ *Mike Blackwell | Technical Analyst, Distribution Services/Rollout Management | RR Donnelley* 1750 Wallace Ave | St Charles, IL 60174-3401 Office: 630.313.7818 mike.blackw...@rrd.com http://www.rrdonnelley.com http://www.rrdonnelley.com/ * mike.blackw...@rrd.com* On Tue, May 19, 2015 at 12:45 PM, Bruno Harbulot br...@distributedmatter.net wrote: On Tue, May 19, 2015 at 6:15 PM, Mike Blackwell mike.blackw...@rrd.com wrote: A Google search suggests Oracle 9.x supports a unary '?' operator (fuzzy match), so the use of '?' in an operator name is not without precedent. Interesting. Do you have any specific link? I'm probably not using the right Google search, but the nearest reference I've found is for Oracle 10, and it seems to use the tilde (~) operator for fuzzy matching: http://www.oracle.com/technetwork/search/oses/overview/new-query-features-in-10-1-8-2-1-132287.pdf Best wishes, Bruno.
Re: [HACKERS] a few thoughts on the schedule
On 2015-05-19 09:43:54 -0700, Joshua D. Drake wrote: On 05/18/2015 08:52 PM, Andres Freund wrote: Maybe we should forget them and just have monthly 'judgefests' where some poor sod summarizes the current state and direction, and we then collaboratively discuss whether we see things going anywhere and if not, what would need to happen that they do. And have a policy that older patches should be preferred over newer ones; but at the same time cull patches continually sitting at the tail end as 'not interesting'. I don't think this will be a productive solution. I would argue that any solution we come up with, somebody is going to think they got the short end of the stick. There will be someone that thinks it is inefficient, that it doesn't suit their needs or that it doesn't work in their paradigm. That is why we don't have a proper issue/bug tracker. That is why we are constantly inventing here instead of relying on the work of others (when it comes to this particular problem). What does that have to do with the suggestion above? That seems entirely unrelated to changing CFs to a different format. I don't know what the solution is but I know I like the idea of a tree freeze except for bug fixes for at least 3 weeks but I would be jumping for joy if we froze the tree except for bug fixes for 6 or 12 weeks. We've done that for pretty much every release so far? I don't care about 9.6 at this point. But you don't develop things for it, so you're in a very different position. It takes a *lot* of time to come up with a serious proposal for a new feature, and then lots more time to come up with a reasonable patch. To get a serious feature into 9.6 you pretty much have to already have started by now. We move so fast anyway, most people I know haven't even migrated to 9.4.x and even more are happily plugging away on 9.2. I don't think that's really related to moving fast. It's just that existing systems don't necessarily need to move - after all they could put the system into production at their respective version. That's different to when you consider adopting/extending postgres for a new use case/product. And there people quit regularly lament a couple problems in postgres. Say if we, and there's been serious talk about that, addressed vacuuming being so painful, that'd certainly increase adoption in the mid term. Greetings, Andres Freund -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On Tue, May 19, 2015 at 6:15 PM, Mike Blackwell mike.blackw...@rrd.com wrote: A Google search suggests Oracle 9.x supports a unary '?' operator (fuzzy match), so the use of '?' in an operator name is not without precedent. Interesting. Do you have any specific link? I'm probably not using the right Google search, but the nearest reference I've found is for Oracle 10, and it seems to use the tilde (~) operator for fuzzy matching: http://www.oracle.com/technetwork/search/oses/overview/new-query-features-in-10-1-8-2-1-132287.pdf Best wishes, Bruno.
Re: [HACKERS] Per row status during INSERT .. ON CONFLICT UPDATE?
On Tue, May 19, 2015 at 10:49 AM, Andres Freund and...@anarazel.de wrote: The RETURNING clause just allows us to return columns, but am unable to find a way to know 'what' happened to a given row. There previously has been discussion about extending RETURNING to allow to return the before/after row. But to me that's a mostly independent feature to ON CONFLICT. That's certainly something we talked about. It could probably be done with some kind of magical expression. I have to wonder how many of the people that are sure that they need this really do, though. Is it really natural to care about this distinction with idiomatic usage? -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] a few thoughts on the schedule
On Tue, May 19, 2015 at 10:35 AM, Andres Freund and...@anarazel.de wrote: I'm not sure. ISTM that a painfull couple hours every now and then are much less bad than the continuous CF we had lately. I personally also find it frustrating to go through the CF and see a good portion of things that I never can see going anywhere, but that still suck up resources. I'd actually be willing to do triage every now and then; but I don't think it should always be the same person. For one it does come with power, for another it's nice to now always be the person having to tell people that their stuff isn't relevant/good/whatever enough. It's also not good to needlessly build up SPOFs. It's pretty hard to tell someone that they're working on something that doesn't matter to us. That's why it happens comparatively rarely. If I told some new contributor that their patch was not worth our time, I'd fully expect some other experienced hacker to show up 5 minutes later and tell me I'm wrong, unless perhaps the idea was shockingly bad, which is rare. Unless talented reviewers can get such job offers, we are going to continue to have trouble making ends meet. Hasn't every talented reviewer gotten job offers shortly afterwards in the last few years? The ones that accept don't necessarily work that much in the community, but several seem to. And I think in the case of several people the reason they don't, is less the company, but that it's emotionally draining. I think that's very true, and often unacknowledged. Reviewing other people's work can be very difficult. I do not enjoy conflict with other people on this mailing list one bit, and that's getting harder to deal with on a personal level over time, not easier. -- Peter Geoghegan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] INSERT ... ON CONFLICT DO UPDATE with _any_ constraint
On 19 May 2015 at 11:49, Geoff Winkless pgsqlad...@geoff.dj wrote: On 19 May 2015 at 16:32, I wrote: In the event that the INSERT triggers a constraint that the UPDATE fails to resolve, it will still fail in exactly the same way that running the ON CONFLICT on a specific constraint would fail, so it's not like you gain any extra value from specifying the constraint, is it? I don't know why I wrote this paragraph, it's just the product of me thinking of something else at the same time: UPDATE obviously doesn't resolve a conflict as such. Thinking about it more, I suppose if multiple constraints end up triggering for the same INSERT, it would require UPDATEs of multiple rows. Is that the issue? I'm sure we'll be asked these questions many times. Can you comment on whether the docs are sufficiently detailed to explain this answer? -- Simon Riggshttp://www.2ndQuadrant.com/ http://www.2ndquadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training Services
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On Tue, May 19, 2015 at 7:51 PM, Greg Sabino Mullane g...@turnstep.com wrote: -BEGIN PGP SIGNED MESSAGE- Hash: RIPEMD160 Dave Cramer opined: It would seem that choosing ? for operators was ill advised; I'm not convinced that deprecating them is a bad idea. If we start now, in 5 years they should be all but gone Ha ha ha ha ha! That's a good one. We still have clients on Postgres 7! Five years is way too short to replace something that major. I think deprecation doesn't necessarily imply removal. It seems that the two operators could exist together by creating a second operator with the same characteristics as suggested by Frank Heikens in this post: http://stackoverflow.com/questions/27573778/postgresql-jsonb-and-jdbc/27580137#27580137 This would also make it easier to backport these operators into existing installations (even on 9.4), thereby making the transition easier. I don't know enough about PostgreSQL's implementation, but I presume this is effectively just giving an alias for the same operation, and hopefully, the query engine could benefit from indices created using either notations interchangeably. (This is probably the most important feature when changing one notation for another.) In addition, the argument regarding the time it can take users to upgrade works both ways. If I understood correctly from your message yesterday, you've only implemented the latest workaround using \? in DBD::Pg quite recently, which would equally require users to be able to upgrade to a more recent version of DBD::Pg (or PHP/PDO where the workaround doesn't seem to be implemented at all yet). Admittedly, I guess it might often be easier to upgrade the client side than the database server, but I'm not sure that is always the case (some frontends can potentially be awkward to update, whereas a database upgrade can be smoother... It varies...). Best wishes, Bruno.
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On 19 May 2015 at 15:02, Tom Lane t...@sss.pgh.pa.us wrote: Greg Sabino Mullane g...@turnstep.com writes: Dave Cramer opined: It would seem that choosing ? for operators was ill advised; I'm not convinced that deprecating them is a bad idea. If we start now, in 5 years they should be all but gone Ha ha ha ha ha! That's a good one. We still have clients on Postgres 7! Five years is way too short to replace something that major. Yeah, that's a big problem for this line of thought. Even if we had consensus today, the first release that would actually contain alternative operators would be 9.6, more than a year out (since 9.5 is past feature freeze now). It would take several years after that before there would be any prospect of removing the old ones, and several years more before PG versions containing the old operators were out of support. Now there are different ways you could look at this. From the perspective of a particular end user, you could imagine instituting a shop policy of not using the operators containing '?' as soon as you had a release where there were alternatives. So in that context you might have a fix available as soon as 9.6 came out. But from the perspective of a driver author who has to support queries written by other people, the problem would not be gone for at least ten years more. Changing the driver's behavior sounds like a more practical solution. The current JDBC driver doesn't really support anything beyond 8.4 except for CRUD operations. We are also are no longer supporting JVM's older than 1.6 in the current driver. People who insist on staying on old code get what they get. I don't see a problem with saying after a certain date we just don't support it in the current code. After all I have heard rumblings about deprecating V2 protocol ? FWIW, I was content to leave this alone. JDBC has a workable solution. However I've not seen a good argument for continuing to use the ? operator as it's conflicts with many clients and is apparently not in the SQL standard. Dave Cramer dave.cramer(at)credativ(dot)ca http://www.credativ.ca
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
-BEGIN PGP SIGNED MESSAGE-
Hash: RIPEMD160
I did find some alternatives discussed a couple of years back, like
{postgres qm} and operator(?); the later simply being to allow the
operator to be quoted inside operator()
Yes, we (DBD::Pg) looked at using at some of the JDBC-ish alternatives
like the (very verbose) vendor escape clauses, but settled on the simplicity of
a single backslash in the end. See part of the discussion here:
http://www.nntp.perl.org/group/perl.dbi.users/2014/12/msg37057.html
- --
Greg Sabino Mullane g...@turnstep.com
End Point Corporation http://www.endpoint.com/
PGP Key: 0x14964AC8 201505191520
http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8
-BEGIN PGP SIGNATURE-
iEYEAREDAAYFAlVbjQQACgkQvJuQZxSWSsgYhACfUfztfxZBQEwESqRYkfRco29M
pAUAoO9qA5IWN96UXsh9iASspiEYfAfF
=k8Gl
-END PGP SIGNATURE-
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] RFC: Non-user-resettable SET SESSION AUTHORISATION
On 19/05/15 20:46, Andres Freund wrote: On 2015-05-19 14:41:06 -0400, Robert Haas wrote: On Tue, May 19, 2015 at 12:29 PM, Andres Freund and...@anarazel.de wrote: On 2015-05-19 10:53:10 -0400, Robert Haas wrote: That seems like a kludge to me. If the cookie leaks out somhow, which it will, then it'll be insecure. I think the way to do this is with a protocol extension that poolers can enable on request. Then they can just refuse to forward any reset authorization packets they get from their client. There's no backward-compatibility break because the pooler can know, from the server version, whether the server is new enough to support the new protocol messages. That sounds like a worse approach to me. Don't you just need to hide the session authorization bit in a function serverside to circumvent that? I'm apparently confused. There's nothing you can do to maintain security against someone who can load C code into the server. I must be misunderstanding you. It very well might be me that's confused. But what's stopping a user from doing a RESET SESSION AUTHORIZATION; in a DO block or something? I guess you are intending that a RESET SESSION AUTHORIZATION is only allowed on a protocol level when the protocol extension is in use? If I understand Robert correctly, he was talking about setting and resetting this on protocol level (with the assistance of pooler) so there is no way to circumvent that from SQL no matter how you mask the command. I think that idea is quite sound. -- Petr Jelinek http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training Services -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On Tue, May 19, 2015 at 8:04 PM, Greg Sabino Mullane g...@turnstep.com wrote: -BEGIN PGP SIGNED MESSAGE- Hash: RIPEMD160 Bruno Harbulot asked for a devil's advocate by saying: My main point was that this is not specific to JDBC. Considering that even PostgreSQL's own ECPG is affected, the issue goes probably deeper than it seems. I'm just not convinced that passing the problem onto connectors, libraries and ultimately application developers is the right thing to do here. Well, one could argue that it *is* their problem, as they should be using the standard Postgres way for placeholders, which is $1, $2, $3... As I was saying in another message on this thread a few hours ago, it appears that ? is reserved for placeholders for Dynamic SQL according to the SQL specifications, and that would be exactly what ECPG is using as far as I understand. Recommending that all drivers implement \? as a semi-standard workaround is actually a much more difficult problem than it seems: it requires following the development of each project, making the case to each community (assuming they're all open source), and reasonable in-depth knowledge of their respective implementation, also assuming that \? won't cause further problems there (of course, all that is easier if you're already working on that particular project). That's actually where we are right now. And it's not really our job to make the case to each community - it is the responsibility of each project to solve the problem, presumably because of pressure from their users. ... except if those communities made the assumption that ? was indeed reserved for placeholders according to the SQL specifications. (I might have misinterpreted where that part of the spec is applicable, since I can't claim I've absorbed the entire set of documents.) Even according to what you're saying this issue has required a first workaround back in 2008, and another one earlier this year, probably due to concerns that weren't spotted when implementing the first workaround (this also presumably requires users to run a fairly recent version of this connector now). True enough regarding the two changes. But the system worked well, in that someone had a problem, raised a bug, and it got fixed. I'm not sure I see the point about requiring recent versions of the connector - that's true for lots of bug fixes and features. This one at least is fairly optional with many existing workarounds (e.g. use $1, quote things in a different way). This model of development also requires the users to be able to upgrade their connectors to a recent release, which may also affect other dependencies (depending on the complexity of the overall system). Best wishes, Bruno.
Re: [HACKERS] a few thoughts on the schedule
On 18 May 2015 at 23:34, Robert Haas robertmh...@gmail.com wrote: On May 18, 2015, at 10:41 PM, Andres Freund and...@anarazel.de wrote: On 2015-05-19 11:34:49 +0900, Michael Paquier wrote: +1 for moving it at least 1 month. 2015-06-15 also collides with pgcon, which probably isn't the best idea. I do think we should try hard doing a triage at the start of a CF and not many with experience in the project are going to have time around then. So, to where do we move it? We probably need to schedule at least the first CF now. Just to 2015-07-15? That'd leave us enough room to schedule the rest at pgcon. Honestly, that seems awful soon. I would have thought maybe August 15th. We have this discussion every year, but I would like to skip that. +1 to 2015-07-15 and then the same schedule as this release. Constant fine tuning the dates doesn't really help, it just creates the impression that discussion might make the dates flexible which works against us, even though I might otherwise agree with them. That allows us to release in Sept, without conflicting with CFs. I suggest we go Beta1 on June 1, so we can discuss any problems arising in person in Ottawa. It's quicker than normal, but if we've lost a month or two we should just skip the usual open items chase, which can be done in parallel with users finding and reporting real bugs. -- Simon Riggshttp://www.2ndQuadrant.com/ http://www.2ndquadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training Services
Re: [HACKERS] Problems with question marks in operators (JDBC, ECPG, ...)
On May 19, 2015 07:04:56 PM Greg Sabino Mullane wrote: Bruno Harbulot asked for a devil's advocate by saying: My main point was that this is not specific to JDBC. Considering that even PostgreSQL's own ECPG is affected, the issue goes probably deeper than it seems. I'm just not convinced that passing the problem onto connectors, libraries and ultimately application developers is the right thing to do here. Well, one could argue that it *is* their problem, as they should be using the standard Postgres way for placeholders, which is $1, $2, $3... Shirley you are joking: Many products use JDBC as an abstraction layer facilitating (mostly) seamless switching between databases. I know the product I worked on did. Are you advocating that every single statement should use SELECT * FROM foo WHERE bar = $1 on pg and SELECT * FROM foo WHERE bar = ? on every other database? A database is only as valuable as the the part of the outside world it can interact with. Large parts of the data-consuming world are developed in java using JDBC. If your opinion is that JDBC developers should adapt themselves to pg then you instantaneously diminish the value of pg. jan -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
