On Sat, Sep 10, 2016 at 8:33 PM, Robert Haas wrote:
> On Sat, Sep 10, 2016 at 3:19 AM, Michael Paquier
> wrote:
>> On Fri, Sep 9, 2016 at 4:01 PM, Kuntal Ghosh
>> wrote:
> - If WAL consistency check is enabled
On Sat, Sep 10, 2016 at 12:49 PM, Michael Paquier
wrote:
> On Fri, Sep 9, 2016 at 4:01 PM, Kuntal Ghosh
> wrote:
>
- In recovery tests (src/test/recovery/t), I've added wal_consistency
parameter in the existing scripts. This
On Sun, Sep 11, 2016 at 12:39 AM, Andres Freund wrote:
> On 2016-09-10 17:23:21 +0530, Amit Kapila wrote:
>> >
>>
>> I may be missing something here, but why would it contend on a lock,
>> as per locking scheme proposed by Alvaro, access to sequence object
>> will need a share
On Sun, Sep 11, 2016 at 4:10 AM, Mark Kirkwood
wrote:
>
>
> performed several 10 hour runs on size 100 database using 32 and 64 clients.
> For the last run I rebuilt with assertions enabled. No hangs or assertion
> failures.
>
Thanks for verification. Do you think
On 9/10/16, Jim Nasby wrote:
> On 9/3/16 6:01 PM, Gavin Flower wrote:
>> I am curious as to the use cases for other possibilities.
>
> A hex or base64 type might be interesting, should those ever get created...
> --
> Jim Nasby, Data Architect, Blue Treble Consulting,
On 9/6/16 5:18 AM, Craig Ringer wrote:
I think something automatic that we clearly define as unstable and not
to be relied upon would be preferable. Plus we already have much of the
infrastructure in elog.c as used by errcontext etc.
Actually, I wish this was a straight-up logging level
On 9/3/16 6:01 PM, Gavin Flower wrote:
I am curious as to the use cases for other possibilities.
A hex or base64 type might be interesting, should those ever get created...
--
Jim Nasby, Data Architect, Blue Treble Consulting, Austin TX
Experts in Analytics, Data Architecture and PostgreSQL
On 9/9/16 6:29 AM, Dmitry Dolgov wrote:
Regarding to the previous conversations [1], [2], here is a patch (with some
improvements from Nikita Glukhov) for the generic type subscription.
Awesome! Please make sure to add it to the Commit Fest app.
--
Jim Nasby, Data Architect, Blue Treble
On 8/1/16 11:38 AM, Bruce Momjian wrote:
I am hoping for a "novice" mode that issues warnings about possible
bugs, e.g. unintentionally-correlated subselect, and this could be part
of that.
Somewhat related; I've recently been wondering about a mode that
disallows Const's in queries coming
On 9/8/16 4:55 AM, Emre Hasegeli wrote:
The main problem that has been discussed before was the indexes. One
way is to tackle with it is to reindex all the tables after the
operation. Currently we are doing it when the datatype of indexed
columns change. So it should be possible, but very
Hi,
As I understand it, a merge join will currently read all tuples from both
subqueries (besides early termination). I believe it should be possible to
take advantages of the indexes on one or both of the tables being read from
to skip a large number of tuples that would currently be read. As an
On 9/8/16 11:35 PM, Tom Lane wrote:
This isn't simple because there are often *lots* of variants. You
> don't just want to see the "top 10" candidate plans, because they're
> probably a bunch of small variants on the same plan; the ones you'll
> be interested in will probably be very different
V2 of this patch:
Changes:
* rebased to most recent master
* removed non-tap test that assumed the existence of Unix sed program
* added non-tap test that assumes the existence of perl
* switched from filename/program to filename/is_program to more closely
follow patterns in copy.c
* slight
On 09/09/16 14:50, Mark Kirkwood wrote:
Yeah, good suggestion about replacing (essentially) all the indexes
with hash ones and testing. I did some short runs with this type of
schema yesterday (actually to get a feel for if hash performance vs
btree was compareable - does seem tp be) - but
Christian Convey writes:
>If that's correct, then it sounds like the only way Joy's commit has
>a chance of acceptance is if Peter's commit is rejected.
>Because Peter's commit might be merged as part of the 2016-09
>commitfest, but Joy's can show up
Hi Heikki,
Could I ask you a newbie-reviewer question about something I'm seeing
here? https://commitfest.postgresql.org/10/776/
>From some reading I've done (e.g., Stephen Frost's PGCon 2011 slides),
I got the impression that a successful patch would always have this
sequence of states in
>
>
> Without having at least compared EXPLAIN outputs from the two boxes, you
> have no business jumping to that conclusion.
>
> If EXPLAIN does show different plans, my first instinct would be to wonder
> whether the pg_stats data is equally up-to-date on both boxes.
>
>
> Thanks. It sounds like worst-case scenario, I perform an unneeded
> review. I'll give it a shot.
Hi guys,
Apologies for more boring process-related questions, but any pointers
would be greatly appreciated...
I'm a bit confused about how PG's code-review process is meant to
handle this C++
Robins Tharakan writes:
> I completely agree. With 'irrelevant' I was only trying to imply that
> irrespective of the complexity of the query, a replicated box was seeing
> similar slowness whereas a Restored DB wasn't. It felt that the SQL itself
> isn't to blame here...
On Fri, 9 Sep 2016 at 09:39 Andres Freund wrote:
> On 2016-09-07 23:37:31 +, Robins Tharakan wrote:
> > If someone asks for I could provide SQL + EXPLAIN, but it feels
> irrelevant
> > here.
>
> Why is that? information_schema are normal sql queries, and some of them
>
I wrote:
> I looked into this a little. There are at least three things we could
> do here:
> 1. Decide that showing walsenders is a good thing. I'm not really
> sure why it isn't -- for example, we could take the trouble to display
> the current replication command as the walsender's activity.
On Thu, Sep 8, 2016 at 10:40 AM, Roger Pack wrote:
> My apologies if this was already requested before...
>
> I think it would be fantastic if postgres had an "explain the explain"
> option:
> Today's explain tells us what loops and scans were used, and relative
> costs,
On Sat, Sep 10, 2016 at 12:04 AM, Heikki Linnakangas wrote:
>> Did I misunderstand something? I'm applying the first patch of Peter's
>> series (cap number of tapes), then your first one (remove prefetch)
>> and second one (use larger read buffers).
>
>
> Yes. I have been testing
On 2016-09-10 17:23:21 +0530, Amit Kapila wrote:
> On Thu, Sep 1, 2016 at 12:00 AM, Andres Freund wrote:
> > On 2016-08-31 14:23:41 -0400, Tom Lane wrote:
> >> Andres Freund writes:
> >> > On 2016-08-31 13:59:48 -0400, Tom Lane wrote:
> >> >> You are
>Personally I wouldn't be very happy about an IEEE754 assumption.
Ok, so let's avoid autoconf man. There is no real explanations of the
ground for this assumption, just a reference to paper of David
Goldberg (and there is no metion about IEEE754 is absoulte
everywhere). BTW, may be we can ask
On Fri, Sep 9, 2016 at 5:22 PM, Claudio Freire wrote:
> Since it is true that doing so would make it impossible to keep the
> asserts about tupindex in tuplesort_heap_root_displace, I guess it
> depends on how useful those asserts are (ie: how likely it is that
> those
I wrote:
> Fujii Masao writes:
>> On Sat, Aug 20, 2016 at 6:13 AM, Tom Lane wrote:
>>> Use LEFT JOINs in some system views in case referenced row doesn't exist.
>> This change causes pg_stat_activity to report the "bogus" information about
>>
Michael Paquier writes:
> Thanks. I was a bit too lazy to look at the history to get this piece
> of history out... And so the code that is presently in the MSVC
> scripts should have been updated at the same time as those
> compilations have been relaxed, but it got
Michael Paquier writes:
> Thanks. I was a bit too lazy to look at the history to get this piece
> of history out... And so the code that is presently in the MSVC
> scripts should have been updated at the same time as those
> compilations have been relaxed, but it got
> Looking for and improving test coverage for RLS is a good suggestion,
> but let's not link the fate of the issue reported here with this
> requirement. I have spent some time looking at this patch and this
> looks in rather good shape to me (you even remembered to use the
> prefix regress_* for
Mattia writes:
> attached is a patch which adds support to localized month names in
> to_date() and to_timestamp() functions.
Seems like a fine goal.
> I thought about reusing from_char_seq_search() but localized month
> names use different capitalization according to the
On Sat, Sep 10, 2016 at 3:19 AM, Michael Paquier
wrote:
> On Fri, Sep 9, 2016 at 4:01 PM, Kuntal Ghosh
> wrote:
- If WAL consistency check is enabled for a rmgrID, we always include
the backup image in the WAL record.
>>>
>>> What
Michael Paquier writes:
> On Fri, Sep 9, 2016 at 11:39 PM, Tom Lane wrote:
>> So this change would deal nicely with the "peer application on the remote
>> host is suddenly stopped" case, at the price of being not nice about any
>> of the other
On Thu, Sep 1, 2016 at 12:00 AM, Andres Freund wrote:
> On 2016-08-31 14:23:41 -0400, Tom Lane wrote:
>> Andres Freund writes:
>> > On 2016-08-31 13:59:48 -0400, Tom Lane wrote:
>> >> You are ignoring the performance costs associated with eating 100x more
On Tue, Aug 30, 2016 at 8:01 AM, Andres Freund wrote:
> On 2016-08-30 07:57:19 +0530, Amit Kapila wrote:
>> I will write such a test case either in this week or early next week.
>
> Great.
>
Okay, attached patch adds some simple tests for pg_stat_statements.
One thing to note
On 8/22/16 10:28 AM, Alvaro Herrera wrote:
> Peter Eisentraut wrote:
>
>> I'm not happy that utils/acl.h has prototypes for aclchk.c, because
>> acl.h is included all over the place. Perhaps I should make a
>> src/include/catalog/aclchk.c to clean that up.
>
> I've been bothered by that too in
On 9/5/16 10:35 PM, Tom Lane wrote:
> In this viewpoint, we'd keep the sequence-specific data in a pg_sequence
> catalog. pg_sequence rows would be extensions of the associated pg_class
> rows in much the same way that pg_index rows extend the pg_class entries
> for indexes. We should supply a
On Wed, Aug 24, 2016 at 3:31 AM, Michael Paquier
wrote:
> On Tue, Aug 23, 2016 at 10:50 PM, Amit Kapila
> wrote:
> > On Tue, Aug 23, 2016 at 6:11 PM, Michael Paquier
> > wrote:
> >> On Tue, Aug 23, 2016 at 8:02 PM,
On 3 Sep. 2016 9:22 pm, "Michael Paquier" wrote:
>
> On Sat, Sep 3, 2016 at 12:42 AM, Magnus Hagander
wrote:
> > On Fri, Sep 2, 2016 at 8:50 AM, Michael Paquier <
michael.paqu...@gmail.com>
> > wrote:
> >> On Fri, Sep 2, 2016 at 2:20 AM, Peter
Review of 0004-Make-libpqwalreceiver-reentrant.patch:
This looks like a good change.
typo: _PG_walreceirver_conn_init
For libpqrcv_create_slot(), slotname should be const char *.
Similarly, for slotname in libpqrcv_startstreaming*() and conninfo in
libpqrcv_connect(). (the latter two
Hi,
attached is a patch which adds support to localized month names in
to_date() and to_timestamp() functions.
The patch is fairly simple but I want to discuss the approach and
implementation:
Using the TM modifier as in to_char() was already discussed some years
ago:
On 9/3/16 9:23 AM, Michael Paquier wrote:
> On Sat, Sep 3, 2016 at 10:22 PM, Michael Paquier
> wrote:
>> On Sat, Sep 3, 2016 at 10:21 PM, Michael Paquier
>> wrote:
>>> Oh, well. I have just implemented it on top of the two other patches
>>>
On Fri, Sep 9, 2016 at 10:18 PM, David Steele wrote:
> On 9/6/16 10:25 PM, Michael Paquier wrote:
>>
>>
>> On Wed, Sep 7, 2016 at 12:16 AM, David Steele wrote:
>>
>>> Attached is a new patch that adds sgml documentation. I can expand on
>>> each
>>>
On Fri, Sep 9, 2016 at 11:39 PM, Tom Lane wrote:
> Haribabu Kommi writes:
>> On Thu, Jun 2, 2016 at 6:51 PM, Kyotaro HORIGUCHI <
>> horiguchi.kyot...@lab.ntt.co.jp> wrote:
>>> After a process termination without PQfinish() of a client,
>>> server
On Fri, Sep 9, 2016 at 4:01 PM, Kuntal Ghosh wrote:
>>> - If WAL consistency check is enabled for a rmgrID, we always include
>>> the backup image in the WAL record.
>>
>> What happens if wal_consistency has different settings on a standby
>> and its master? If for
On 09/10/2016 04:21 AM, Claudio Freire wrote:
On Fri, Sep 9, 2016 at 9:51 PM, Claudio Freire wrote:
On Fri, Sep 9, 2016 at 8:13 AM, Heikki Linnakangas wrote:
Claudio, if you could also repeat the tests you ran on Peter's patch set on
the other
On Thu, Sep 8, 2016 at 10:04 PM, Tom Lane wrote:
> The core code has never used xslt at all.
Yes.
> Some quick digging in the git
> history suggests that contrib/xml2 wasn't very clean about this before
> 2008:
> [...]
> Both of those fixes postdate our native-Windows port,
On Sat, Sep 10, 2016 at 3:55 AM, Adam Brightwell
wrote:
>> Perhaps we should extend rowsecurity test with a more comprehensive
>> set of tests rather than just fix the COPY one?
>
> I think more tests that provide value are always a *good* thing,
> however, would
48 matches
Mail list logo