On Sat, May 26, 2018 at 02:10:52PM +1200, Thomas Munro wrote:
> I also wondered about this when trying to figure out how to write a
> TAP test for recovery testing with tablespaces, for my undo proposal.
> I was starting to wonder about either allowing relative paths or
> supporting some kind of va
On Sat, May 26, 2018 at 9:17 AM, Ashwin Agrawal wrote:
> To generate uniqueness for the path between primary and standby need to use
> something which is not represented within database. So will be random to
> some degree. Like one can use PORT number of postmaster. As only need to
> generate uniq
On 05/25/18 21:16, Andrew Gierth wrote:
>> "Tom" == Tom Lane writes:
> Tom> about that for v11, but I'd favor trying to improve the situation
> Tom> in v12.
>
> Yeah. Another issue I ran into is that if you use SPI_prepare_params,
> then you have to use SPI_execute_plan_with_paramlist, it's
> "Tom" == Tom Lane writes:
>>> /*
>>> * GAH. To do parameter type checking properly, we have to install our
>>> * own global post-parse hook transiently.
>>> */
>> Gah, indeed. Thanks for the heads up. I would never have guessed it'd
>> be that fiddly.
Tom> Yikes. That seems pretty
Chapman Flack writes:
> On 05/25/18 20:07, Andrew Gierth wrote:
>> /*
>> * GAH. To do parameter type checking properly, we have to install our
>> * own global post-parse hook transiently.
>> */
>> ...
>> PG_TRY();
>> {
>> pllua_spi_prev_parse_hook = post_parse_analyze_hook;
>> post_parse_analyze_h
On 05/25/18 20:07, Andrew Gierth wrote:
> >> with writing a ParserSetupHook that's just a thin wrapper around
> >> parse_variable_parameters ?
>
> That's what I did in pllua-ng. The tricky bit was in arranging to also
> call check_variable_parameters; I considered skipping that part, but
> that
> "Tom" == Tom Lane writes:
> Chapman Flack writes:
>> As for bringing it along to the modern API, am I on the right track
>> with writing a ParserSetupHook that's just a thin wrapper around
>> parse_variable_parameters ?
Tom> Seems reasonable from here.
That's what I did in pllua-ng.
"David G. Johnston" writes:
> On Fri, May 25, 2018 at 3:33 PM, Tom Lane wrote:
>> Even if you'd made a case why we should consider them equal,
>> those would be very good reasons not to change behavior that's
>> stood for 17 years.
> This is true, and the alternative doesn't have the supporting
I got this error message via -fsanitized=undefined:
jsonfuncs.c:5169:12: runtime error: load of value 127, which is not a
valid value for type '_Bool'
The query was
select ts_headline('{}'::jsonb, tsquery('aaa & bbb'));
This calls the C function ts_headline_jsonb_byid_opt(), which calls
transfo
On Fri, May 25, 2018 at 06:24:07PM +0300, Heikki Linnakangas wrote:
> On 25 May 2018 17:44:16 EEST, Robert Haas wrote:
>> It seems to me that this is really another sort of thing altogether.
>> Whether or not you want to insist on channel binding is a completely
>> separate thing from which channe
Chapman Flack writes:
> As for bringing it along to the modern API, am I on the right track
> with writing a ParserSetupHook that's just a thin wrapper around
> parse_variable_parameters ?
Seems reasonable from here.
regards, tom lane
On Fri, May 25, 2018 at 3:33 PM, Tom Lane wrote:
> Alexey Bashtanov writes:
> > Comparison of timetz values looks a bit weird to me, as
> > '22:00+02'::timetz > '21:00+01'::timetz.
>
> Perhaps, but I don't think there's a reasonable case for considering
> them equal, either. In the other places
On 05/25/2018 06:01 PM, Tom Lane wrote:
> Offhand I don't believe SPI exposed a way to do that before 9bedd128d.
> Does it matter? Pre-9.0 releases are long out of support by now,
> which means they're full of known data-loss hazards and security bugs.
Well, my exploration arose from discovering
Alexey Bashtanov writes:
> Comparison of timetz values looks a bit weird to me, as
> '22:00+02'::timetz > '21:00+01'::timetz.
Perhaps, but I don't think there's a reasonable case for considering
them equal, either. In the other places where obviously-different
values compare equal, such as zero
Hello,
Comparison of timetz values looks a bit weird to me, as
'22:00+02'::timetz > '21:00+01'::timetz.
I can see this behavior introduced by commit 2792374c in 2001
with the comment "timetz was just plain broken (some possible pairs of
values were neither < nor = nor >)".
The in-code commen
On 2018-05-25 17:47:37 -0400, Tom Lane wrote:
> Andres Freund writes:
> > Moving discussion to -hackers. Tom, I think you worked most with this
> > code, your input would be appreciated.
>
> Yeah, the assumption in the relcache is that the only part of a nailed
> catalog's relcache entry that re
Chapman Flack writes:
> Am I on the right track here? Is what I'm looking to do something
> that became possible in SPI in 9.0 and wasn't before, or did I overlook
> a way it could have been done pre-9.0 ?
Offhand I don't believe SPI exposed a way to do that before 9bedd128d.
Does it matter? Pre
On Fri, May 25, 2018 at 05:17:12PM -0400, Tom Lane wrote:
> Robert Haas writes:
> > On Fri, May 25, 2018 at 1:53 PM, Amit Langote
> > wrote:
> >> Seems here that we call find_appinfos_by_relids here for *all*
> >> partitions, even if all but one partition may have been pruned. I
> >> haven't st
Andres Freund writes:
> Moving discussion to -hackers. Tom, I think you worked most with this
> code, your input would be appreciated.
Yeah, the assumption in the relcache is that the only part of a nailed
catalog's relcache entry that really needs to be updated intrasession is
the relfilenode m
> "Joe" == Joe Conway writes:
Joe> My colleague Yogesh Sharma discovered an assert in nested SQL
Joe> procedure calls after ROLLBACK is used. Minimal test case and
Joe> backtrace below. I have not yet tried to figure out exactly what
Joe> is going on beyond seeing that it occurs in pg_pla
> On May 25, 2018, at 5:17 PM, Tom Lane wrote:
>
> Maybe it's all right to decide that this rejiggering can be left
> for v12 ... did we promise anyone that it's now sane to use thousands
> of partitions?
Per beta release, we’ve only said “improved SELECT query performance due
to enhanced parti
On 05/24/2018 11:26 PM, Heikki Linnakangas wrote:
> Cool stuff!
>
> On 25/05/18 06:35, Andres Freund wrote:
>> For example, this converts this converts TPCH's Q01:
+1
Wicked cool!
--
Crunchy Data - http://crunchydata.com
PostgreSQL Support for Secure Enterprises
Consulting, Training, & Open So
On Fri, May 25, 2018 at 7:33 AM, Tom Lane wrote:
> Ashwin Agrawal writes:
> > Proposing to create directory with timestamp at time of creating
> tablespace
> > and create symbolic link to it instead.
>
> I'm skeptical that this solves your problem. What happens when the CREATE
> TABLESPACE comm
Robert Haas writes:
> On Fri, May 25, 2018 at 1:53 PM, Amit Langote wrote:
>> Seems here that we call find_appinfos_by_relids here for *all*
>> partitions, even if all but one partition may have been pruned. I
>> haven't studied this code in detail, but I suspect it might be
>> unnecessary, alth
My colleague Yogesh Sharma discovered an assert in nested SQL procedure
calls after ROLLBACK is used. Minimal test case and backtrace below. I
have not yet tried to figure out exactly what is going on beyond seeing
that it occurs in pg_plan_query() where the comment says "Planner must
have a snapsh
Hi,
Moving discussion to -hackers. Tom, I think you worked most with this
code, your input would be appreciated.
Original discussion is around:
http://archives.postgresql.org/message-id/20180524211311.tnswfnjwnii54htx%40alvherre.pgsql
On 2018-05-24 17:13:11 -0400, Alvaro Herrera wrote:
> On 201
On 05/24/2018 02:30 AM, Chapman Flack wrote:
> In 9.0, there's SPI_prepare_params, which seems promising; it accepts
> an arbitrary ParserSetupHook "to control the parsing of external parameter
> references." But its documentation doesn't suggest what to use as the
> ParserSetupHook to say "please
On Fri, May 25, 2018 at 1:53 PM, Amit Langote wrote:
> Seems here that we call find_appinfos_by_relids here for *all*
> partitions, even if all but one partition may have been pruned. I
> haven't studied this code in detail, but I suspect it might be
> unnecessary, although I might be wrong.
Ugg
On Sun, May 6, 2018 at 11:53:34AM -0400, Tom Lane wrote:
> What sort of changes do we get if we remove those two flags as you prefer?
> It'd help to see some examples.
>
> Since we just went to a new perltidy version, and made some other
> policy changes for it, in HEAD, it'd make sense to make a
Robbie Harwood writes:
> Thomas Munro writes:
>
>> On Thu, May 24, 2018 at 8:00 AM, Robbie Harwood wrote:
>>
>>> Zombie patch is back from the dead.
>>
>> Hi Robbie,
>>
>> Robots[1] vs zombies:
>>
>> + $postgres->RemoveFile('src/backennd/libpq/be-gssapi-common.c');
>>
>> Typo, breaks on Windows
Mike Palmiotto writes:
> On 05/24/2018 03:30 PM, Tom Lane wrote:
>> I tried to run the regression test for sepgsql on F28 (so I could
>> fix the now-obsolete expected-file therein). It fails at this
>> preparatory step:
>> $ sudo semodule -u sepgsql-regtest.pp
> This error appears to be due to a
On Fri, May 25, 2018 at 11:49 PM, Robert Haas wrote:
> On Fri, May 25, 2018 at 10:30 AM, Thomas Reiss
> wrote:
>> Then I used the following to compare the planning time :
>> explain (analyze) SELECT * FROM t1 WHERE dt = '2018-05-25';
>>
>> With PostgreSQL 10, planning time is 66ms, in v11, plann
Le 25/05/2018 à 16:49, Robert Haas a écrit :
> On Fri, May 25, 2018 at 10:30 AM, Thomas Reiss
> wrote:
>> Then I used the following to compare the planning time :
>> explain (analyze) SELECT * FROM t1 WHERE dt = '2018-05-25';
>>
>> With PostgreSQL 10, planning time is 66ms, in v11, planning ris
On Fri, May 25, 2018 at 12:14 PM, Greg Clough wrote:
> Many thanks for the quick consideration, even if it's ultimately a rejection.
> Figuring out some SQL that will work across all platforms, versions, and
> compile-time options will be "fun", but I'm up for a challenge.
Why would you need d
I notice there are still several places in pl_exec.c like this:
case SPI_ERROR_TRANSACTION:
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot begin/end transactions in PL/pgSQL"),
errhint("Use a
>> On Fri, May 25, 2018 at 10:11 AM, Andrew Dunstan
>> wrote:
>>> He's proposing an extra column to show the actual value used, so
>>> distinguishing them should be a problem.
>
>> For most settings, that column would just be a duplicate. For a
>> handful, it would pull in the value of some other
Robert Haas writes:
> On Fri, May 25, 2018 at 11:21 AM, Tom Lane wrote:
>> Looking at what mod_stmt is used for, we've got
>> (1) the Assert that's crashing, and its siblings, which are just meant
>> to cross-check that mod_stmt is consistent with the SPI return code.
>> (2) two places that decid
On Fri, May 25, 2018 at 11:21 AM, Tom Lane wrote:
> Robert Haas writes:
>> For reasons that I'm not quite sure about, the following test case
>> crashes in v11, but not earlier versions:
>
> Crashes just fine in prior versions for me, at least as far back as 9.3,
> but probably much further. Not
On Fri, May 18, 2018 at 11:21 AM, Masahiko Sawada wrote:
> Regarding to API design, should we use 2PC for a distributed
> transaction if both two or more 2PC-capable foreign servers and
> 2PC-non-capable foreign server are involved with it? Or should we end
> up with an error? the 2PC-non-capable
On 25 May 2018 17:44:16 EEST, Robert Haas wrote:
>On Wed, May 23, 2018 at 2:46 AM, Heikki Linnakangas
>wrote:
>> We could provide "tls-unique" and "tls-server-end-point" in addition
>to
>> those, but I'd consider those to be developer only settings, useful
>only for
>> testing the protocol.
>
>
Robert Haas writes:
> For reasons that I'm not quite sure about, the following test case
> crashes in v11, but not earlier versions:
Crashes just fine in prior versions for me, at least as far back as 9.3,
but probably much further. Note that I was doing an extra select fun()
right after creatin
Got it, will go for second method. Will let you guys know the progress.
2018-05-25 4:05 GMT-07:00 Aleksander Alekseev :
> Hello Charles,
>
> I suggest to begin with the second approach and add to_jsonb/from_jsonb
> later. Both approaches are fine but it seems to me that most users would
> expect
Thanks for correcting me, will definitely study citext and see how a new
type is registered in plugin.
2018-05-24 23:23 GMT-07:00 Aleksandr Parfenov :
> On Thu, 24 May 2018 18:25:28 -0700
> Charles Cui wrote:
> > The second is to provide thrift type just like json or jsonb. When you
> > create
On Fri, May 25, 2018 at 10:30 AM, Thomas Reiss wrote:
> Then I used the following to compare the planning time :
> explain (analyze) SELECT * FROM t1 WHERE dt = '2018-05-25';
>
> With PostgreSQL 10, planning time is 66ms, in v11, planning rise to
> 143ms. I also did a little test with more than 20
On Wed, May 23, 2018 at 2:46 AM, Heikki Linnakangas wrote:
> "tls-unique" and "tls-server-end-point" are overly technical to users. They
> don't care which one is used, there's no difference in security.
> Furthermore, if we add another channel binding type in the future, perhaps
> because someone
Chris Bandy writes:
> On 5/24/18 2:31 PM, Tom Lane wrote:
>> Andrew Gierth writes:
>>> There's also the option of adding an explicit function
>>> date_trunc(text,date) returns date, which is a workaround that I (and
>>> probably quite a few other people) have used.
> Are we in agreement that the
Ashwin Agrawal writes:
> Proposing to create directory with timestamp at time of creating tablespace
> and create symbolic link to it instead.
I'm skeptical that this solves your problem. What happens when the CREATE
TABLESPACE command is replicated to the standby with sub-second delay?
Clock sk
On Fri, May 25, 2018 at 10:22 AM, Tom Lane wrote:
> Personally, what I'd rather do is try to get rid of GUC behaviors like
> "the effective value depends on something else". But convenience and
> backwards compatibility may be arguments against that.
Yeah. The dependency between various GUCs is
On 5/24/18 2:31 PM, Tom Lane wrote:
Andrew Gierth writes:
"Tom" == Tom Lane writes:
Tom> Yeah. There are two relevant variants of date_trunc():
[...]
Tom> So we probably ought to change the docs here.
There's also the option of adding an explicit function
date_trunc(text,date) return
Hello,
I spent some time to test the new features on partitioning with the
beta1. I noticed a potentially huge performance regression with
plan-time partition pruning.
To show the issue, I used this DO statement to generate some partitions,
one per day :
DO $$
DECLARE
part_date date;
ddl text
On Wed, May 23, 2018 at 6:03 AM, Laurenz Albe wrote:
> I think the cache should be limited in size, perhaps by work_mem.
> Also, there probably should be a GUC for this, defaulting to "off".
Eh, why? Generally we want performance optimizations turned on by
default; otherwise only highly-knowledg
On Wed, May 23, 2018 at 12:51 PM, Tom Lane wrote:
> Ah. That would work, though it'd make the number of subquery executions
> even less predictable (since some logically-equal values would compare
> as physically unequal).
In most cases that seems fine. It might not be fine with the subquery
co
Robert Haas writes:
> On Fri, May 25, 2018 at 10:11 AM, Andrew Dunstan
> wrote:
>> He's proposing an extra column to show the actual value used, so
>> distinguishing them should be a problem.
> For most settings, that column would just be a duplicate. For a
> handful, it would pull in the value
On Fri, May 25, 2018 at 10:11 AM, Andrew Dunstan
wrote:
> He's proposing an extra column to show the actual value used, so
> distinguishing them should be a problem.
For most settings, that column would just be a duplicate. For a
handful, it would pull in the value of some other GUC. If somebod
On Fri, May 25, 2018 at 9:58 AM, Tom Lane wrote:
> Greg Clough writes:
>> I would like to propose that we expose the "active" value of parameters in
>> pg_settings, instead of "-1". In this example below, when it's set to "-1"
>> I need to know that autovacuum_work_mem is related to the settin
Kyotaro HORIGUCHI writes:
> It is set for other kinds of message, (parse, bind, execute). I
> think fastpath, close, flush and sync don't need that. If it is
> reasonable to assume that we can see debug_query_string in the
> DESCRIBE path, the attached patch would work.
I think this patch is a ba
For reasons that I'm not quite sure about, the following test case
crashes in v11, but not earlier versions:
create table abc(n int);
create table xyz(n int);
create function fun() returns int as $$begin insert into abc values
(1); return 1; end$$ language plpgsql;
create or replace rule rule1 a
Greg Clough writes:
> I would like to propose that we expose the "active" value of parameters in
> pg_settings, instead of "-1". In this example below, when it's set to "-1" I
> need to know that autovacuum_work_mem is related to the setting of
> maintenance_work_mem, so that I can determine t
Hi Hackers,
I would like to propose that we expose the "active" value of parameters in
pg_settings, instead of "-1". In this example below, when it's set to "-1" I
need to know that autovacuum_work_mem is related to the setting of
maintenance_work_mem, so that I can determine that the actual s
On 25 May 2018 at 13:12, Magnus Hagander wrote:
>
>
> On Fri, May 25, 2018 at 7:28 AM, Michael Paquier
> wrote:
>>
>> Hi all,
>>
>> When attempting to use multiple times pg_replication_slot_advance on a
>> slot, then the caller gets back directly InvalidXLogRecPtr as result,
>> for example:
>> =#
On Fri, May 25, 2018 at 7:28 AM, Michael Paquier
wrote:
> Hi all,
>
> When attempting to use multiple times pg_replication_slot_advance on a
> slot, then the caller gets back directly InvalidXLogRecPtr as result,
> for example:
> =# select * from pg_replication_slot_advance('popo', 'FF/0');
> sl
Hello Hackers,
This propose a way to develop "Table-level" Transparent Data Encryption (TDE)
and Key Management Service (KMS) support in
PostgreSQL.
Issues on data encryption of PostgreSQL
==
Currently, in PostgreSQL, data encryption can be using pgcrypto Tool.
However, it is inconvenie
Hello Charles,
I suggest to begin with the second approach and add to_jsonb/from_jsonb
later. Both approaches are fine but it seems to me that most users would
expect a separate type thus it's more important.
--
Best regards,
Aleksander Alekseev
signature.asc
Description: PGP signature
Currently, if primary and standby are setup on same machine (which is
always the case for development), CREATE TABLESPACE xyz LOCATION '/abc',
primary and mirror both write to "/abc/TABLESPACE_VERSION_DIRECTORY"
directory. Collision is certainly not an issue in any production deployment
but seems s
Thank you for the comment.
At Fri, 25 May 2018 09:05:21 +0300, Heikki Linnakangas wrote
in <466a3c6d-7986-8cb1-d908-e85aa6a09...@iki.fi>
> On 25/05/18 07:45, Kyotaro HORIGUCHI wrote:
> > Hello.
> > I happened to see the following in XLogWrite.
> >
> >> ereport(PANIC,
> >> (errcode_for_f
On Thu, May 24, 2018 at 11:50 PM, Andres Freund wrote:
> On 2018-05-25 09:40:10 +0300, Heikki Linnakangas wrote:
> > On 25/05/18 09:25, Asim Praveen wrote:
> > > My parochial vision of the overhead is restricted to 4 * NBuffers of
> > > additional shared memory, as 4 bytes are being added to Buff
Michael Paquier writes:
> but for now I would recommend to register this patch to the next
> commit fest under the category "Bug Fixes" so as it does not fall into
> the cracks: https://commitfest.postgresql.org/18/
>
> I have added an entry to the open items in the section for older bugs:
> htt
On Fri, May 25, 2018 at 01:19:58PM +0900, Kyotaro HORIGUCHI wrote:
> The case is not of an empty file. read() reads 0 bytes without
> error while lseek have told that the file has *more* data. I
> don't think that can happen. How about just commenting with
> something like the following?
Actually
68 matches
Mail list logo
