Re: Finalizing logical replication limitations as well as potential features

On 05/01/18 05:35, Joshua D. Drake wrote: On 01/04/2018 01:26 PM, Alvaro Herrera wrote: Joshua D. Drake wrote: We just queue/audit the changes as they happen and sync up the changes after the initial sync completes. This already happens.  There is an initial sync, and there's logical

Re: FP16 Support?

Just for your information. I tried to implement "float2" according to IEEE 754 specification, as a custom data type of PG-Strom. https://github.com/heterodb/pg-strom/blob/master/src/float2.c https://github.com/heterodb/pg-strom/blob/master/sql/float2.sql The recent GPU device (Maxwell or later)

Re: [HACKERS] Parallel tuplesort (for parallel B-Tree index creation)

On Sat, Jan 6, 2018 at 3:47 AM, Peter Geoghegan wrote: > On Tue, Jan 2, 2018 at 8:43 PM, Rushabh Lathia > wrote: > > I agree that plan_create_index_workers() needs to count the leader as a > > normal worker for the CREATE INDEX. So what you proposing is

Re: [HACKERS] taking stdbool.h into use

On Sat, Dec 30, 2017 at 10:31:45AM -0500, Tom Lane wrote: > Surely bool and bool8 should have identical Datum representations, > so I'm not sure they need different DatumGet/GetDatum macros. [... refresh update ...] Yeah sure. I am a bit disturbed by the fact that DatumGetBool() enforces a case

Re: refactor subscription tests to use PostgresNode's wait_for_catchup

On Mon, Jan 08, 2018 at 09:46:21PM -0500, Peter Eisentraut wrote: > It appears that we have unwittingly created some duplicate and > copy-and-paste-prone code in src/test/subscription/ to wait for a > replication subscriber to catch up, when we already have > almost-sufficient code in PostgresNode

Re: [HACKERS] Proposal: Local indexes for partitioned table

On 9 January 2018 at 09:54, Alvaro Herrera wrote: > I removed the pg_index.indparentidx column that previous versions add, > and replaced it with pg_inherits rows. This makes the code a little bit > bulkier in a couple of places, but nothing terrible. As a benefit, >

Re: [HACKERS] Replication status in logical replication

On Sun, Jan 7, 2018 at 7:50 PM, Simon Riggs wrote: > On 26 December 2017 at 00:26, Masahiko Sawada wrote: >> On Tue, Dec 26, 2017 at 1:10 AM, Petr Jelinek >> wrote: >>> On 21/11/17 22:06, Masahiko Sawada wrote:

Re: BUG #14941: Vacuum crashes

On Thu, Dec 28, 2017 at 10:46:18PM +, Bossart, Nathan wrote: > I agree, this makes more sense. I've made this change in v3 of 0003. Based on the opinions gathered on this thread, 0001 and 0002 seem to be in the shape wanted, and those look good for shipping. Now for 0003 we are not there

Re: Parallel append plan instability/randomness

On Tue, Jan 9, 2018 at 12:48 AM, Tom Lane wrote: > Robert Haas writes: >> On Sun, Jan 7, 2018 at 11:40 PM, Amit Kapila wrote: >>> One theory that can explain above failure is that the costs of >>> scanning some of the sub-paths

Re: Condition variable live lock

On Mon, Jan 8, 2018 at 7:02 PM, Tom Lane wrote: > I'm not real sure BTW why we have some callers that ereport and some > that just exit(1). Seems like it would be better to be consistent, > though I'm not entirely sure which behavior to standardize on. I think at one point

Re: Invalid pg_upgrade error message during live check

On Mon, Jan 8, 2018 at 06:26:21PM -0500, Bruce Momjian wrote: > On Mon, Jan 8, 2018 at 10:43:00AM -0500, Robert Haas wrote: > > On Fri, Jan 5, 2018 at 9:11 PM, Bruce Momjian wrote: > > > On Fri, Jan 5, 2018 at 04:58:39PM -0500, Bruce Momjian wrote: > > >> Pg_upgrade is able

Incorrect comment for expand_single_inheritance_child

Here is part of comments for expand_single_inheritance_child: * expand_single_inheritance_child * Expand a single inheritance child, if needed. * * If this is a temp table of another backend, we'll return without doing * anything at all. Otherwise, build a RangeTblEntry and an

Use of term hostaddrs for multiple hostaddr values

Hi all, While looking at the documentation of libpq, I have noticed that the term hostaddrs is used to qualify multiple values of hostaddr. This looks incorrect to me, as this is not the name of a connection parameter. Please find attached a patch to address this inconsistency. One error message

Re: Comment typo in postgres_fdw.c

(2017/12/28 4:03), Robert Haas wrote: On Wed, Dec 27, 2017 at 4:35 AM, Etsuro Fujita wrote: Here is a small patch for fixing $Subject: s/it's/its/. Committed. Thanks! Best regards, Etsuro Fujita

refactor subscription tests to use PostgresNode's wait_for_catchup

It appears that we have unwittingly created some duplicate and copy-and-paste-prone code in src/test/subscription/ to wait for a replication subscriber to catch up, when we already have almost-sufficient code in PostgresNode to do that more compactly. So I propose this patch to consolidate that.

Re: PATCH: logical_work_mem and logical streaming of large in-progress transactions

On 1/3/18 15:13, Tomas Vondra wrote: > Forgot to mention that the v4 also extends the CREATE SUBSCRIPTION to > allow customizing the streaming and memory limit. So you can do > > CREATE SUBSCRIPTION ... WITH (streaming=on, work_mem=1024) > > and this subscription will allow streaming, and

Re: PATCH: logical_work_mem and logical streaming of large in-progress transactions

On 1/3/18 14:53, Tomas Vondra wrote: >> I don't see the need to tie this setting to maintenance_work_mem. >> maintenance_work_mem is often set to very large values, which could >> then have undesirable side effects on this use. > > Well, we need to pick some default value, and we can either use

Re: proposal: alternative psql commands quit and exit

Everaldo Canuto wrote: > Change my patch will make psql different from other sql clients I use > (sqlplus and mysql). Well, I don't think we're too hot about copying their behavior exactly. This thread discussed the behavior at length and a consensus was found after much thinking. No one is

Re: portal pinning

On 1/8/18 15:27, Andrew Dunstan wrote: > This seems like a good idea, and the code change is tiny and clean. I > don't know of any third party PLs or other libraries might be pinning > the portals already on their own. How would they be affected if they did? They would get an error if they tried

Re: Enhance pg_stat_wal_receiver view to display connected host

On Fri, Jan 05, 2018 at 09:15:36AM -0300, Alvaro Herrera wrote: > Haribabu Kommi wrote: > > > And also not returning "default host" details, because for the conninfo > > without any host details, the return value must be NULL. But this change > > may break the backward compatibility of the

Re: Unimpressed with pg_attribute_always_inline

Hi, On 2018-01-08 20:09:27 -0500, Peter Eisentraut wrote: > On 1/8/18 19:56, Tom Lane wrote: > > Peter Geoghegan writes: > >> Anyway, ISTM that it should be possible to make > >> pg_attribute_always_inline have no effect in typical debug builds. > >> Wouldn't that make everyone

Re: [HACKERS] Transactions involving multiple postgres foreign servers

On Mon, Jan 1, 2018 at 7:12 PM, Ashutosh Bapat wrote: > On Thu, Dec 28, 2017 at 11:08 AM, Masahiko Sawada > wrote: >> >>> (1) >>> Why don't you use the existing global variable MyXactFlags instead of the >>> new TransactionDidWrite? Or,

Re: Unimpressed with pg_attribute_always_inline

Peter Geoghegan writes: > Anyway, ISTM that it should be possible to make > pg_attribute_always_inline have no effect in typical debug builds. > Wouldn't that make everyone happy? That would improve matters, but do we have access to the -O switch level as an #if condition? The

Re: Unimpressed with pg_attribute_always_inline

On Mon, Jan 8, 2018 at 4:12 PM, Tom Lane wrote: > When I complained that always_inline inhibits debuggability, I did NOT > mean what shows up in perf reports. I'm talking about whether you can > break at, or single-step through, a function reliably and whether gdb > knows

Re: Unimpressed with pg_attribute_always_inline

On 2018-01-08 19:12:08 -0500, Tom Lane wrote: > Andres Freund writes: > > Unless this pg_attribute_always_inline gets a lot more widely > > proliferated I don't see a need to change anything. Debuggability isn't > > meaningfully impacted by seing more runtime attributed to > >

Re: Condition variable live lock

Thomas Munro writes: > On Sun, Jan 7, 2018 at 10:00 AM, Tom Lane wrote: >> Another loose end that I'm seeing here is that while a process waiting on >> a condition variable will respond to a cancel or die interrupt, it will >> not notice

Re: Condition variable live lock

Thomas Munro writes: > On Sun, Jan 7, 2018 at 10:00 AM, Tom Lane wrote: >> I began with the intention of making no non-cosmetic changes, but then >> I started to wonder why ConditionVariablePrepareToSleep bothers with a >> !proclist_contains

Re: Unimpressed with pg_attribute_always_inline

On 2018-01-05 00:11:19 +1300, Thomas Munro wrote: > On Tue, Jan 2, 2018 at 4:58 PM, Thomas Munro > wrote: > > On Tue, Jan 2, 2018 at 4:17 PM, Tom Lane wrote: > >> gaur | nodeHashjoin.c:167: warning: `always_inline' attribute > >>

Re: [HACKERS] [PATCH] Generic type subscripting

Hi, On 2018-01-04 16:47:50 +0100, Dmitry Dolgov wrote: > diff --git a/src/backend/executor/execExpr.c b/src/backend/executor/execExpr.c > index 16f908037c..ee50fda4ef 100644 > --- a/src/backend/executor/execExpr.c > +++ b/src/backend/executor/execExpr.c > @@ -64,7 +64,8 @@ static void

Re: Invalid pg_upgrade error message during live check

On Mon, Jan 8, 2018 at 10:43:00AM -0500, Robert Haas wrote: > On Fri, Jan 5, 2018 at 9:11 PM, Bruce Momjian wrote: > > On Fri, Jan 5, 2018 at 04:58:39PM -0500, Bruce Momjian wrote: > >> Pg_upgrade is able to run in --check mode when the old server is still > >> running.

Re: Parallel append plan instability/randomness

I wrote: >> In the regression test case at hand, the startup costs are all zero so >> this change wouldn't improve the test case's stability. So I'm thinking >> that in addition, it would be a good idea for these functions to break >> exact compare_path_costs ties in some arbitrary but

Re: update portal-related memory context names and API

Peter Eisentraut wrote: > Subject: [PATCH 1/2] Update portal-related memory context names and API > > Rename PortalMemory to TopPortalContext, to avoid confusion with > PortalContext and align naming with similar top-level memory contexts. > > Rename PortalData's "heap" field to portalContext.

Re: [HACKERS] Proposal: Local indexes for partitioned table

Hi Alvaro, On 01/08/2018 03:36 PM, Alvaro Herrera wrote: Jesper Pedersen wrote: Maybe a warning for existing indexes on the same column(s), but with a different type, should be emitted during ATTACH, e.g. [detach one partition, replace index with a different one, attach partition] Of

Re: [HACKERS] Proposal: Local indexes for partitioned table

Jesper Pedersen wrote: > Maybe a warning for existing indexes on the same column(s), but with a > different type, should be emitted during ATTACH, e.g. > [detach one partition, replace index with a different one, attach > partition] > Of course, this could also fall under index maintenance and

Re: update portal-related memory context names and API

On 12/19/2017 12:31 PM, Peter Eisentraut wrote: > ISTM that some of the portal-related memory context naming is a bit > antiquated and at odds with current terminology. In this patch, I > propose to rename PortalMemory to TopPortalContext and rename > Portal->heap to Portal->portalContext, and

Re: [HACKERS] VACUUM and ANALYZE disagreeing on what reltuples means

On 01/08/2018 08:39 PM, Tom Lane wrote: > Tomas Vondra writes: >> As I already mentioned, Tom's updated patch is better than what I >> posted initially, and I agree with his approach to the remaining >> issues he pointed out. But I somehow assumed that he's already

Re: Condition variable live lock

On Sun, Jan 7, 2018 at 6:38 PM, Tom Lane wrote: >> Actually ... perhaps a better design would be to have >> ConditionVariable[PrepareTo]Sleep auto-cancel any prepared sleep for >> a different condition variable, analogously to what we just did in >> ConditionVariableBroadcast,

Re: Parallel append plan instability/randomness

On Mon, Jan 8, 2018 at 1:26 PM, Jim Finnerty wrote: > I see. Maybe the EXPLAIN output can be made more stable when "EXPLAIN (costs > FALSE)", though? I think that would be fixing the problem from the wrong end. We want to actually get a stable choice of plan, not display

Re: Parallel append plan instability/randomness

I wrote: > In the regression test case at hand, the startup costs are all zero so > this change wouldn't improve the test case's stability. So I'm thinking > that in addition, it would be a good idea for these functions to break > exact compare_path_costs ties in some arbitrary but deterministic

Re: [HACKERS] Proposal: Local indexes for partitioned table

Hi Alvaro, On 01/04/2018 09:30 AM, Alvaro Herrera wrote: v11 fixes the dependency problem I mentioned in https://postgr.es/m/20171229203818.pqxf2cyl4g2wre6h@alvherre.pgsql and adds some test to cover that stuff. Thanks, make check-world passes. Maybe a warning for existing indexes on the

Re: PL/Python SD dict wiped?

On 1/8/18 11:59, Ken Huffman wrote: > I'm fine with per-session initializing of prepared statements, but is > there PL/Python mechanism that spans sessions to minimize ConfigParser > calls? Nothing built-in. You would have to use a separate external storage mechanism of some kind. It could be a

Re: [HACKERS] [PATCH] Incremental sort

On Mon, Jan 8, 2018 at 2:29 PM, Antonin Houska wrote: > Alexander Korotkov wrote: > > > Antonin Houska wrote: > > > > Shouldn't the test contain *both* cases? > > > Thank you for pointing that. Sure, both cases are better. I've

Re: pgbench - add \if support

Hello Dmitry, Thanks for working on this. I had just a quick look at it, but I hope I'll have time to post a proper review. In the meantime I'm wondering what am I doing wrong here (I see a similar example in your first message)? ``` -- test.sql \if random(0, 99) < 85 \set test 1 \else

Re: [HACKERS] VACUUM and ANALYZE disagreeing on what reltuples means

On 01/05/2018 05:25 AM, Stephen Frost wrote: > Tomas, > > * Michael Paquier (michael.paqu...@gmail.com) wrote: >> On Sun, Nov 19, 2017 at 6:40 AM, Tomas Vondra >> wrote: >>> Thanks for looking into this. I agree your patch is more consistent and >>> generally

Re: Parallel append plan instability/randomness

Robert Haas writes: > [ example where current parallel-append dispatch algorithm loses ] I should clarify my thinking a bit more: I was imagining that we might switch to a system where, as each worker comes free, it makes a dynamic choice of which subplan to take next.

Re: [HACKERS] WIP: Covering + unique indexes.

Hello! The patch does not apply currently. Anastasia, can you, please, rebase the patch? Best regards, Andrey Borodin.

Re: [HACKERS] GSoC 2017 : Patch for predicate locking in Gist index

On 8 January 2018 at 22:44, Shubham Barai wrote: > > > On 5 January 2018 at 03:18, Alexander Korotkov > wrote: > >> On Thu, Jan 4, 2018 at 7:07 PM, Andrey Borodin >> wrote: >> >>> 29 нояб. 2017 г., в 22:50, Shubham

Re: pgbench - add \if support

> On 4 January 2018 at 07:32, Fabien COELHO wrote: > > Another rebase to try to please the patch tester. Thanks for working on this. I had just a quick look at it, but I hope I'll have time to post a proper review. In the meantime I'm wondering what am I doing wrong here (I

Re: Doc tweak for huge_pages?

On Fri, Dec 1, 2017 at 10:09 PM, Thomas Munro wrote: >> On 11/30/17 23:35, Thomas Munro wrote: >>> Hmm. Yeah, it does, but apparently it's not so transparent. So if we >>> mention that we'd better indicate in the same paragraph that you >>> probably don't actually

Re: [HACKERS] [PATCH] Generic type subscripting

> On 7 January 2018 at 23:39, Tom Lane wrote: >> Dmitry Dolgov <9erthali...@gmail.com> writes: >>> On 4 January 2018 at 03:05, Tom Lane wrote: >>> I wonder what happened to the plan to separate assignment and fetch into two >>> different node types. I can

Re: [HACKERS] GSoC 2017 : Patch for predicate locking in Gist index

On 5 January 2018 at 03:18, Alexander Korotkov wrote: > On Thu, Jan 4, 2018 at 7:07 PM, Andrey Borodin > wrote: > >> 29 нояб. 2017 г., в 22:50, Shubham Barai >> написал(а): >> >> I have fixed formatting style. Please

Re: [PROPOSAL] Shared Ispell dictionaries

Hi Arthur, Sorry for the delay, I somehow missed this thread ... On 12/27/2017 10:20 AM, Arthur Zakirov wrote: > On Tue, Dec 26, 2017 at 07:03:48PM +0100, Pavel Stehule wrote: >> >> Tomas had some workable patches related to this topic >> > > Tomas, have you planned to propose it? > I believe

RE: PL/Python SD dict wiped?

I'm fine with per-session initializing of prepared statements, but is there PL/Python mechanism that spans sessions to minimize ConfigParser calls? -Original Message- From: Tom Lane [mailto:t...@sss.pgh.pa.us] Sent: Monday, January 08, 2018 10:23 AM To: Ken Huffman

Re: Parallel append plan instability/randomness

Robert Haas writes: > On Mon, Jan 8, 2018 at 11:42 AM, Tom Lane wrote: >> More generally, I wonder if >> it wouldn't be better to implement this behavior at runtime rather >> than plan time. > Ignoring some details around partial vs. non-partial plans,

Re: [HACKERS] Refactoring identifier checks to consistently use strcmp

> On 08 Jan 2018, at 17:27, Robert Haas wrote: > > On Sat, Jan 6, 2018 at 7:38 PM, Tom Lane wrote: >> It's definitely concerning that the submitted patch introduced a new bug, >> but we have seldom taken the position that bugs in an initial submission

Re: Parallel append plan instability/randomness

Jim Finnerty writes: > Ordering the plan output elements by estimated cost will cause spurious plan > changes to be reported after table cardinalities change. Can we choose an > explain output order that is stable to changes in cardinality, please? I found the code that's

Re: Parallel append plan instability/randomness

On Mon, Jan 8, 2018 at 11:28 AM, Jim Finnerty wrote: > Ordering the plan output elements by estimated cost will cause spurious plan > changes to be reported after table cardinalities change. Can we choose an > explain output order that is stable to changes in cardinality,

Re: [HACKERS] [PROPOSAL] Temporal query processing with range types

On Sat, Jan 6, 2018 at 3:29 PM, Simon Riggs wrote: > I'm not very keen on adopting new syntax that isn't in the > SQLStandard. They have a bad habit of doing something completely > different. So a flexible approach will allow us to have functionality > now and we can adopt

Re: Parallel append plan instability/randomness

On Sun, Jan 7, 2018 at 11:40 PM, Amit Kapila wrote: > One theory that can explain above failure is that the costs of > scanning some of the sub-paths is very close due to which sometimes > the results can vary. If that is the case, then probably using > fuzz_factor in

Re: Invalid pg_upgrade error message during live check

On Fri, Jan 5, 2018 at 9:11 PM, Bruce Momjian wrote: > On Fri, Jan 5, 2018 at 04:58:39PM -0500, Bruce Momjian wrote: >> Pg_upgrade is able to run in --check mode when the old server is still >> running. Unfortunately, all supported versions of pg_upgrade generate >> an

Re: PL/Python SD dict wiped?

Ken Huffman writes: > But if I quit and restart psql and do a subsequent db update, the SD dict is > empty again and the one-time logic has to run again. What is clearing the > dict? Is that expected? You've got a new backend process. That dict (or any other Python

PL/Python SD dict wiped?

Is there a known issue with the PL/Python extension clearing the SD dict under some circumstances? As suggested by the documentation, my plpythonu uses the SD dict for one-time (prepare and ConfigParser) logic: CREATE OR REPLACE FUNCTION change_notify() RETURN TRIGGER AS $$ ... If len(SD) == 0:

Re: Condition variable live lock

Thomas Munro writes: > One very small thing after another look: > - Assert(cv_sleep_target == NULL); > + if (cv_sleep_target != NULL) > + ConditionVariableCancelSleep(); > The test for cv_sleep_target != NULL is redundant since >

Re: [HACKERS] UPDATE of partition key

On 4 January 2018 at 02:52, David Rowley wrote: > I'll try to look at the tests tomorrow and also do some testing. I've made a pass over the tests. Again, sometimes I'm probably a bit pedantic. The reason for that is that the tests are not that easy to follow.

Re: [HACKERS] Timeline ID in backup_label file

On Sat, Jan 06, 2018 at 09:56:03AM +0900, Michael Paquier wrote: > Previous patch was incorrect, this was the same v2 as upthread. > Attached is the correct v3. For the archive's sake, this version of the patch has been pushed as 6271fce. Thanks Simon for the commit, and David for the review! --

Re: pgsql: Implement channel binding tls-server-end-point for SCRAM

On Sat, Jan 06, 2018 at 09:10:51AM +0900, Michael Paquier wrote: > On Sat, Jan 6, 2018 at 2:56 AM, Alvaro Herrera > wrote: > Good idea as a whole, but I don't think this is the right approach. As > we include $(bindir) in PATH when running the prove command, why not >

Re: Condition variable live lock

On Mon, Jan 8, 2018 at 5:25 PM, Thomas Munro wrote: > On Sun, Jan 7, 2018 at 10:43 AM, Tom Lane wrote: >> Actually ... perhaps a better design would be to have >> ConditionVariable[PrepareTo]Sleep auto-cancel any prepared sleep for >> a

Re: Parallel append plan instability/randomness

On Mon, Jan 8, 2018 at 2:11 PM, Amit Khandekar wrote: > On 8 January 2018 at 13:35, Amit Kapila wrote: >> On Mon, Jan 8, 2018 at 11:26 AM, Tom Lane wrote: >>> Amit Khandekar writes: The fact that

Re: Parallel append plan instability/randomness

On 8 January 2018 at 13:35, Amit Kapila wrote: > On Mon, Jan 8, 2018 at 11:26 AM, Tom Lane wrote: >> Amit Khandekar writes: >>> The fact that b_star gets moved from 5th position to the first >>> position in the scans,

Re: Parallel append plan instability/randomness

On Mon, Jan 8, 2018 at 11:26 AM, Tom Lane wrote: > Amit Khandekar writes: >> The fact that b_star gets moved from 5th position to the first >> position in the scans, indicates that it's cost shoots up from 1.04 to >> a value greater than 1.16. It does