On Sat, Dec 16, 2017 at 12:57 AM, Andreas Seltenreich
wrote:
> Hi,
>
> sqlsmith just crashed a parallel worker while testing master at
> 699bf7d05c. I can reproduce it with the following recipe on a fresh
> regression database. Backtrace and query plan below as well.
>
This seems to be another
On 13 December 2017 at 00:33, Beena Emerson wrote:
> PFA the updated patch, this can be applied over the v13 patches [1]
> over commit 487a0c1518af2f3ae2d05b7fd23d636d687f28f3
Hi Beena,
Thanks for posting an updated patch.
I've been looking over this and I think that the use of the
PartitionDis
On Tue, Dec 12, 2017 at 2:26 PM Peter Eisentraut <
peter.eisentr...@2ndquadrant.com> wrote:
> On 12/12/17 13:03, Jeremy Finzel wrote:
> > To be clear, what I mean is batch updating a large set of data in small
> > pieces so as to avoid things like lock contention and replication lags.
> > Sometime
Andres Freund writes:
> On 2017-12-16 07:52:41 +0900, Michael Paquier wrote:
>> On Sat, Dec 16, 2017 at 3:13 AM, Christoph Berg
>> wrote:
>>> Re: Tom Lane 2017-12-15 <9616.1513351...@sss.pgh.pa.us>
genbki.pl is hardly our only script that prints its $0 ...
>>> As per
>>> https://tests.repro
On Fri, Dec 15, 2017 at 3:21 PM, Andres Freund wrote:
>> Why not fixing that? Reproducible builds are a trend of these days,
>> and what's proposed here is really simple to make PG more compliant
>> with this purpose in mind.
>
> It's not like $0 instead of a hardcoded name in the header actually
On 2017-12-16 07:52:41 +0900, Michael Paquier wrote:
> On Sat, Dec 16, 2017 at 3:13 AM, Christoph Berg
> wrote:
> > Re: Tom Lane 2017-12-15 <9616.1513351...@sss.pgh.pa.us>
> >> Christoph Berg writes:
> >> > Debian's reproducible builds project has revealed that the full build
> >> > path gets emb
On Sat, Dec 16, 2017 at 3:13 AM, Christoph Berg
wrote:
> Re: Tom Lane 2017-12-15 <9616.1513351...@sss.pgh.pa.us>
>> Christoph Berg writes:
>> > Debian's reproducible builds project has revealed that the full build
>> > path gets embedded into server/catalog/schemapg.h:
>>
>> genbki.pl is hardly o
On Sat, Dec 16, 2017 at 4:49 AM, Gene Selkov wrote:
> I noticed somebody took a good care of seg while I was away for the last 20
> years, and I am extremely grateful for that. I have been using it. In the
> meantime, things have changed and now I am almost clueless about how you
> deal with contr
On Sat, Dec 16, 2017 at 2:39 AM, Peter Eisentraut
wrote:
> On 12/13/17 02:35, Michael Paquier wrote:
>> Patch 0001 is simply removing EventTriggerSupportsGrantObjectType(),
>> but shouldn't we keep it and return an error for objects that have no
>> GRANT support? Returning conditionally true looks
Commit d70cf811, from 2014, promoted an Assert() within
IndexBuildHeapScan() to a "can't happen" elog() error, in order to
detect when a parent tuple cannot be found for some heap-only tuple --
if this happens, then it indicates corruption. I think that we should
make it a full ereport(), with an e
Robert Haas wrote:
> On Fri, Dec 15, 2017 at 4:02 PM, Alvaro Herrera
> wrote:
> > 3. Robert's: use CREATE INDEX ON ONLY , which creates a shell
> >index on parent only (no recursion), followed by CREATE INDEX ON
> >. DETACH is not provided. If you ATTACH an index for a
> >partition
Robert Haas wrote:
> On Thu, Nov 30, 2017 at 7:02 AM, Alvaro Herrera
> wrote:
> > Great question. So you're thinking that the planner might have an
> > interest in knowing what indexes are defined at the parent table level
> > for planning purposes; but for that to actually have any effect we wo
On Fri, Dec 15, 2017 at 4:02 PM, Alvaro Herrera wrote:
> 3. Robert's: use CREATE INDEX ON ONLY , which creates a shell
>index on parent only (no recursion), followed by CREATE INDEX ON
>. DETACH is not provided. If you ATTACH an index for a
>partition that already has one index attac
Alvaro Herrera wrote:
> 3. Robert's: use CREATE INDEX ON ONLY , which creates a shell
>index on parent only (no recursion), followed by CREATE INDEX ON
>. DETACH is not provided. If you ATTACH an index for a
>partition that already has one index attached, then (1) the newly
>atta
Hello Teodor,
It may be good for 't' of 'f' but it seems too free grammar to accept 'tr'
or 'fa' or even 'o' which actually not known to be on or off.
Yes, it really works like that. I tried to make something clearer than
"src/bin/psql/variable.c". Maybe I did not succeed.
Ok, I see. Curre
On Fri, Dec 15, 2017 at 7:58 AM, Robert Haas wrote:
> Reviewing the preparatory patch:
I started another review pass over the main patch, so here are some
comments about that. This is unfortunately not a complete review,
however.
- map = ptr->partition_tupconv_maps[leaf_part_index];
+ map = ptr
On Sat, Dec 16, 2017 at 9:13 AM, Robert Haas wrote:
> On Fri, Dec 15, 2017 at 2:10 PM, Jeff Janes wrote:
>> I had hit on the same change. And was also surprised that it was located
>> where it was. With the change, it uses the parallel plan all the way down
>> to LIMIT 1.
>>
>> With the patch,
Hmm, so I'm now unsure what the actual proposals for handling pg_dump
are. We seem to have the following three proposals:
1. Alvaro: use CREATE INDEX ON ONLY (not recursive ), followed
by CREATE INDEX ON , followed by ALTER INDEX
ATTACH PARTITION . I provide an ALTER INDEX DETACH
PART
Matthew Kelly writes:
> I recently fell afoul of a weird edge case while writing an extension. It
> seems Postgres allows for an ambiguous column reference to oid in the where
> clause when joining to pg_roles. It just arbitrarily chooses pg_roles.oid
> and ignores the conflicting name. Exam
On 12/12/17 10:34, Peter Eisentraut wrote:
> But I also wonder whether we shouldn't automatically pin/unpin portals
> in SPI_cursor_open() and SPI_cursor_close(). This makes sense if you
> consider "pinned" to mean "internally generated". I don't think there
> is a scenario in which user code sho
On Fri, Dec 15, 2017 at 2:10 PM, Jeff Janes wrote:
> I had hit on the same change. And was also surprised that it was located
> where it was. With the change, it uses the parallel plan all the way down
> to LIMIT 1.
>
> With the patch, it still satisfies make check, so if it introduces errors
>
Greetings everyone,
I need a data type to represent genomic positions, which will consist of a
string and a pair of integers with interval logic and access methods. Sort
of like my seg type, but more straightforward.
I noticed somebody took a good care of seg while I was away for the last 20
year
I recently fell afoul of a weird edge case while writing an extension. It
seems Postgres allows for an ambiguous column reference to oid in the where
clause when joining to pg_roles. It just arbitrarily chooses pg_roles.oid and
ignores the conflicting name. Example:
postgres=# CREATE TABLE
Hi,
sqlsmith just crashed a parallel worker while testing master at
699bf7d05c. I can reproduce it with the following recipe on a fresh
regression database. Backtrace and query plan below as well.
regards,
Andreas
--8<---cut here---start->8---
set min_parall
On 2017-12-15 11:15:47 -0800, Peter Geoghegan wrote:
> On Fri, Dec 15, 2017 at 11:03 AM, Peter Geoghegan wrote:
> > The elog(), which was itself upgraded from a simple Assert by commit
> > d70cf811, appears in exactly the same form in 9.3+. Things did change
> > there, but they were kept in sync.
On Fri, Dec 15, 2017 at 11:03 AM, Peter Geoghegan wrote:
> The elog(), which was itself upgraded from a simple Assert by commit
> d70cf811, appears in exactly the same form in 9.3+. Things did change
> there, but they were kept in sync.
BTW, if you're going to do it, I would target the similar er
On Fri, Dec 15, 2017 at 12:40 PM, Peter Eisentraut
wrote:
> On 12/14/17 22:59, Rushabh Lathia wrote:
>> I noted that no_priv_msg and not_owner_msg array been removed
>> and code fitted the code into aclcheck_error(). Actually that
>> makes the code more complex then what it used to be. I would
>
On Thu, Dec 14, 2017 at 5:12 PM, Thomas Munro wrote:
> >
> > This looks like a costing bug. The estimated cost of sorting 416,667
> > estimated tuples in one parallel worker is almost identical to the
> estimated
> > cost of sorting 1,000,000 tuples when parallelism is turned off. Adding
> > so
On Fri, Dec 15, 2017 at 10:58 AM, Andres Freund wrote:
>> I have one minor piece of feedback on the upgrading of assertions to
>> ereport()s with ERRCODE_DATA_CORRUPTION: It would be nice if you could
>> upgrade the raw elog() "can't happen" error within
>> IndexBuildHeapRangeScan() to be an erepo
On 2017-12-15 10:46:05 -0800, Peter Geoghegan wrote:
> On Thu, Dec 14, 2017 at 6:30 PM, Andres Freund wrote:
> > Pushed this way. Moved some more relfrozenxid/relminmxid tests outside
> > of the cutoff changes, polished some error messages.
> >
> >
> > Alvaro, Michael, Peter, and everyone else I'
On 2017-12-15 20:25:22 +0900, Michael Paquier wrote:
> On Fri, Dec 15, 2017 at 11:30 AM, Andres Freund wrote:
> > Alvaro, Michael, Peter, and everyone else I'd greatly appreciate if you
> > could have a look at the backported version, just about everything but
> > v10 had conflicts, some of them n
On Thu, Dec 14, 2017 at 6:30 PM, Andres Freund wrote:
> Pushed this way. Moved some more relfrozenxid/relminmxid tests outside
> of the cutoff changes, polished some error messages.
>
>
> Alvaro, Michael, Peter, and everyone else I'd greatly appreciate if you
> could have a look at the backported
Re: Tom Lane 2017-12-15 <9616.1513351...@sss.pgh.pa.us>
> Christoph Berg writes:
> > Debian's reproducible builds project has revealed that the full build
> > path gets embedded into server/catalog/schemapg.h:
>
> genbki.pl is hardly our only script that prints its $0 ...
As per
https://tests.re
On 15 December 2017 at 17:13, Peter Eisentraut
wrote:
> You need to put
>
> set enable-bracketed-paste on
>
> into ~/.inputrc, then it works.
Hmm, looks like that requires a newer version of readline (v7) than I
have here.
Oh well, if support is already there (albeit unavailable) then I'll
leave
On 12/14/17 22:59, Rushabh Lathia wrote:
> I noted that no_priv_msg and not_owner_msg array been removed
> and code fitted the code into aclcheck_error(). Actually that
> makes the code more complex then what it used to be. I would
> prefer the array rather then code been fitted into the function
On 12/13/17 02:35, Michael Paquier wrote:
> Patch 0001 is simply removing EventTriggerSupportsGrantObjectType(),
> but shouldn't we keep it and return an error for objects that have no
> GRANT support? Returning conditionally true looks like a trap waiting
> to take someone in.
I don't understand
On 12/15/17 11:22, Geoff Winkless wrote:
> It occurred to me the other day while people were talking about
> pasting blocks of text creating problems, especially with tabs, that
> xterm bracketed-paste support (also works in at least putty and
> probably others) that would block curses handling and
Hi
It occurred to me the other day while people were talking about
pasting blocks of text creating problems, especially with tabs, that
xterm bracketed-paste support (also works in at least putty and
probably others) that would block curses handling and just paste as-is
would be a useful (and I'm
Christoph Berg writes:
> Debian's reproducible builds project has revealed that the full build
> path gets embedded into server/catalog/schemapg.h:
genbki.pl is hardly our only script that prints its $0 ...
regards, tom lane
Christoph Berg writes:
> Re: Michael Paquier 2017-12-15
>
>> Why reinventing the wheel when there is already --with-extra-version
>> that you can use for the same purpose?
> That modifies the PG version number as such, as what psql is showing
> on connect. I'd think that is too intrusive.
I'm
2017-12-15 14:47 GMT+01:00 Teodor Sigaev :
> 2) In makeVariableValue():
>>> if (pg_strcasecmp(var->svalue, "null") == 0)
>>> ...
>>> else if (pg_strncasecmp(var->svalue, "true", slen)
>>>
>>> mixing of pg_strcasecmp and pg_strNcasecmp. And, IMHO, result of
>>> pg_strncasecmp("tru", "true", 1) will
Alex,
* Alex Kliukin (al...@hintbits.com) wrote:
> On Fri, Dec 15, 2017, at 14:52, Aleksander Alekseev wrote:
> > Completely agree, this project can be an improvement for Stolon (or
> > Patroni, but I personally never tested or used it, also I got a feeling
> > that Google guys will prefer a proje
On Fri, Dec 15, 2017, at 14:52, Aleksander Alekseev wrote:
> Completely agree, this project can be an improvement for Stolon (or
> Patroni, but I personally never tested or used it, also I got a feeling
> that Google guys will prefer a project that is written in Go). This
> would make much more
Alex,
* Alex Kliukin (al...@hintbits.com) wrote:
> On Fri, Dec 15, 2017, at 14:30, Stephen Frost wrote:
> > * Aleksander Alekseev (a.aleks...@postgrespro.ru) wrote:
> >
> > > Regarding the difficulty of the project - in fact it's not that
> > > difficult. Particularly this project can rely on ext
Hello,
On Fri, Dec 15, 2017, at 14:30, Stephen Frost wrote:
> Aleksander,
>
> * Aleksander Alekseev (a.aleks...@postgrespro.ru) wrote:
>
> > Regarding the difficulty of the project - in fact it's not that
> > difficult. Particularly this project can rely on external tools, e.g.
> > use Consul fo
Aleksander Alekseev writes:
> Hi Stephen,
>
>> synchronous_commit isn't the same as synchronous_standby_names ...
>
> What about synchronous_standby_names? Logical replication ignores it and
> doesn't wait for ack's from replicas or something?
It actually works, see [1]:
The name of a standby s
Hi!
I found out the problem in exposing values of t_bits field from
heap_page_items function.
When the number of attributes in table is multiple of eight, t_bits
column shows double number of bits in which data fields are included.
For example:
# create table tbl(f1 int, f2 int, f3 int, f
Hi Stephen,
> synchronous_commit isn't the same as synchronous_standby_names ...
What about synchronous_standby_names? Logical replication ignores it and
doesn't wait for ack's from replicas or something?
> What might be interesting is seeing if Logical Replication could be
> added to Patroni as
2) In makeVariableValue():
if (pg_strcasecmp(var->svalue, "null") == 0)
...
else if (pg_strncasecmp(var->svalue, "true", slen)
mixing of pg_strcasecmp and pg_strNcasecmp. And, IMHO, result of
pg_strncasecmp("tru", "true", 1) will be 0.
Yep, but it cannot be called like that because slen == str
On 12/14/2017 08:55 PM, Tom Lane wrote:
> Andrew Dunstan writes:
>> On 12/14/2017 08:37 PM, Jordan Deitch wrote:
>>> I am unable to build pgindent as it appears the pearltidy source has
>>> been removed from sourceforge:
>>> https://git.postgresql.org/gitweb/?p=postgresql.git;a=blob_plain;f=src/
Aleksander,
* Aleksander Alekseev (a.aleks...@postgrespro.ru) wrote:
> > HA/fail-over is a very broad topic, with a lot of pieces that need to be
> > done such that I'm not sure it's really viable, but perhaps a precursor
> > project (synchronous logical replication seems like a prereq, no?) would
On Wed, Dec 13, 2017 at 5:18 AM, Amit Khandekar wrote:
> Amit Langote informed me off-list, - along with suggestions for
> changes - that my patch needs a rebase. Attached is the rebased
> version. I have also bumped the patch version number (now v29),
> because this as additional changes, again,
Hi Stephen,
> HA/fail-over is a very broad topic, with a lot of pieces that need to be
> done such that I'm not sure it's really viable, but perhaps a precursor
> project (synchronous logical replication seems like a prereq, no?) would
> make more sense. Or, perhaps, a different piece of the HA q
Aleksander,
* Aleksander Alekseev (a.aleks...@postgrespro.ru) wrote:
> > New entries are certainly welcome and encouraged, just be sure to note
> > them as '2018' when you add it.
>
> I proposed a few ideas:
Thanks!
> * High availability / failover based on logical replication
> * Thrift dataty
Re: Michael Paquier 2017-12-15
> On Fri, Dec 15, 2017 at 7:46 PM, Christoph Berg
> wrote:
> > To be able to identify more easily which package a connected server is
> > coming from, I would like to embed the (Debian) package version in the
> > version() output which is coming from PG_VERSION. It
Stefan,
* Stefan Keller (sfkel...@gmail.com) wrote:
> 2017-12-15 4:14 GMT+01:00 Stephen Frost :
> > Unsurprisingly, we'll need to have an Ideas page again, so I've gone
> > ahead and created one (copying last year's):
>
> What about adding "Learned Index" as project task [*]?
> This type of index
On Fri, Dec 15, 2017 at 7:46 PM, Christoph Berg
wrote:
> To be able to identify more easily which package a connected server is
> coming from, I would like to embed the (Debian) package version in the
> version() output which is coming from PG_VERSION. It is fairly easy to
> do that, but it requir
On Fri, Dec 15, 2017 at 11:30 AM, Andres Freund wrote:
> Alvaro, Michael, Peter, and everyone else I'd greatly appreciate if you
> could have a look at the backported version, just about everything but
> v10 had conflicts, some of them not insubstantial.
I have gone through the backpatched versio
To be able to identify more easily which package a connected server is
coming from, I would like to embed the (Debian) package version in the
version() output which is coming from PG_VERSION. It is fairly easy to
do that, but it requires patching configure(.in):
$ ./configure VENDOR_VERSION="Debia
Hi!
> 15 дек. 2017 г., в 15:03, Stefan Keller написал(а):
> What about adding "Learned Index" as project task [*]?
> This type of index looks promising for certain properties.
I have no high expectation from this idea. But feel that it is necessary at
least to validate the idea. I'd sign up as
Hi,
Debian's reproducible builds project has revealed that the full build
path gets embedded into server/catalog/schemapg.h:
/*-
*
* schemapg.h
*Schema_pg_xxx macros for use by relcache.c
*
* Portions Copyright (c) 1
Hi,
2017-12-15 4:14 GMT+01:00 Stephen Frost :
> Unsurprisingly, we'll need to have an Ideas page again, so I've gone
> ahead and created one (copying last year's):
What about adding "Learned Index" as project task [*]?
This type of index looks promising for certain properties.
:Stefan
[*] "The
Hi Stephen,
> New entries are certainly welcome and encouraged, just be sure to note
> them as '2018' when you add it.
I proposed a few ideas:
* High availability / failover based on logical replication
* Thrift datatype support
Hope these ideas are good enough for GSoC.
--
Best regards,
Alek
On 15.12.2017 01:21, Michael Paquier wrote:
On Fri, Dec 15, 2017 at 6:15 AM, Alvaro Herrera wrote:
Konstantin Knizhnik wrote:
If you still thing that additional 16 bytes per relation in statistic is too
high overhead, then I will also remove autotune.
I think it's pretty clear that these ad
Hello, I noticed while an investigation that pg_prewarm is
forgetting to register a tranche.
Before the second parameter of LWLockRegisterTranche() became
char * in 3761fe3, that would lead to a crash for a
--enable-dtrace build, but currently it not likely.
On the other hand as far as reading a
65 matches
Mail list logo
