Hi,
On Fri, Nov 07, 2025 at 05:18:41PM +, Dagfinn Ilmari Mannsåker wrote:
> Peter Eisentraut writes:
>
> > On 07.11.25 16:03, Bertrand Drouvot wrote:
> >
> >> +#define pg_attribute_deprecated(msg) [[deprecated(msg)]]
> >> +#elif defined(__GNUC__) || defined(__clang__)
> >
> > The __clang__ p
On Sat, Nov 8, 2025 at 7:21 PM Jim Jones wrote:
> > I personally lean towards that maybe we should keep it simple and go
> > back to the origins of the patch which just tried to show if it is a
> > standby or not given how complex it is to tell that we are in read-only
> > or not. The alternative
On Sun, Nov 9, 2025 at 5:21 AM Joshua Shanks wrote:
>
> Hi,
>
> The recent patch for passwordFromFile() error handling highlighted that many
> other out-of-memory error messages in libpq still aren't wrapped with
> libpq_gettext().
>
> Attached is a patch that systematically wraps the remaining
On Wed, Nov 5, 2025 at 6:43 AM Naga Appani wrote:
>
> Understanding
>
> Based on reading the relevant parts of multixact.c and observing the runtime
> behavior, both approaches seem to run into limitations when trying to derive a
> “remaining members” value outside the backend. I may
On Sat, Sep 27, 2025 at 2:28 AM Masahiko Sawada wrote:
>
> On Fri, Sep 26, 2025 at 8:02 AM Álvaro Herrera wrote:
> >
> > On 2025-Sep-26, Chao Li wrote:
> >
> > > I agree to remove the field from AlterPublicationStmt, but I think we
> > > should retain "Assert(!stmt)”. Because Assert() is a way to
Thanks for the clarification. However, I still believe this is out of scope
for the CREATE POLICY DDL. The command ALTER TABLE ... ENABLE ROW LEVEL
SECURITY seems more appropriate as part of the CREATE TABLE reconstruction
rather than CREATE POLICY.
That said, I’m open to adding it if the majority
On Mon, Nov 10, 2025 at 9:41 AM Haruna Miwa wrote:
>
> Hi,
>
> As described in the manual[1], PREPARE ... AS command supports MERGE and
> VALUES statements, but current psql's tab completion does not suggest them.
> I created a patch that enables MERGE and VALUES to appear as tab completion
> su
Hi!
I'm trying to use pg_getaddrinfo_all() with NULL hints, but got segfault.
According to man(3) getaddrinfo, hints may be passed as NULL. In this
case af_family
is equivalent to AF_UNSPEC.
I'm adding a patch to pg_getaddrinfo_all() to make it available by
passing hintp as NULL.
--
With
On Tue, 4 Nov 2025 at 04:28, Alexander Korotkov
wrote:
> I have rechecked this. It appears that pg_createsubscriber writes the
> recovery configuration to the output and only in verbose mode. So,
> it's far no guaranteed that this information would be accessible. One
> may run pg_createsubscri
Hi Sawada-San.
Some review comments for v24-0001.
==
src/backend/replication/slot.c
ReplicationSlotsDropDBSlots:
1.
LWLockAcquire(ReplicationSlotControlLock, LW_SHARED);
for (i = 0; i < max_replication_slots; i++)
{
ReplicationSlot *s;
+ bool invalidated;
char*slotname;
int
On Mon, Nov 10, 2025 at 11:07 AM Chao Li wrote:
> I just did a test. In the test, I inserted a tuple with the same primary key
> so that the inserts fails by the unique key constraint which breaks the
> pipeline, and some random select statements followed. And I added some debug
> messages in d
On Tue, Nov 4, 2025 at 10:23 PM Álvaro Herrera wrote:
>
> On 2025-Nov-04, Peter Smith wrote:
>
> > On Sat, Nov 1, 2025 at 5:02 AM Álvaro Herrera wrote:
>
> > > I'm not opposed to 0001 (to master only), but I think the lines of
> > > dashes are a little excessively noisy. Are there other opinions
On 2025-11-10 03:03, Tom Lane wrote:
Philip Warner writes: On 2025-11-09 04:06, Tom Lane wrote:
IIRC you'd need to use --create to prod pg_dump to produce
a CREATE DATABASE command as well as any database-level
ALTER commands.
Is that a good approach?
Without --create,
pg_dump is not char
On Mon, Nov 10, 2025 at 10:30:31AM +0800, Chao Li wrote:
> Is really confused. The error message says “maximum of 64”, but the
> test right uses a name of length 64. I know that the tricky is the
> ‘\0’ terminator, but should SQL writer have to keep mind about the
> ‘\0’ terminator? Should they jus
Hi,
With a sorted commited.xip array, we could replace the iteration with
two binary searches to find the interval to keep.
Proposed Optimization
-
Use binary search to locate the boundaries of XIDs to remove, then
compact with a single memmove. The key insight requires under
On Sat, Nov 8, 2025 at 8:24 AM Bruce Momjian wrote:
>
>
> Agreed, patch applied, thanks.
>
Thanks Bruce.
--
Best Wishes,
Ashutosh Bapat
On Tue, Oct 14, 2025 at 12:11 AM Tom Lane wrote:
>
> Andrew Dunstan writes:
> > I'm late to the party on this, but I wonder if it wouldn't be better to
> > use a type-neutral parameter name here, like "source", which could cover
> > all these cases, instead of "string", "bytes", etc.
>
> +1 for t
> On Nov 10, 2025, at 09:11, Michael Paquier wrote:
>
> Hi all,
>
> While looking at a recent patch for injection points that has resulted
> in 16a2f706951e, I have been reminded that the point name, library
> name and function name have hardcoded limits, and it is now possible
> to have them
> On Nov 7, 2025, at 17:33, Fujii Masao wrote:
>
> On Fri, Nov 7, 2025 at 9:07 AM Chao Li wrote:
>> If you put this explanation to the inline comment, things would get clearer.
>> But based on this explanation, I just got the other doubt. When a pipeline
>> is aborted, res is NULL, but we s
Hi,
We had some off-list discussions about this patch (me, Daniel and
Andres), and Andres mentioned he suspects the patch may be breaking PITR
in some way. We didn't have any example of that, but PITR seems like a
pretty fundamental feature, so I took it seriously and decided to do
some stress tes
Hi all,
While looking at a recent patch for injection points that has resulted
in 16a2f706951e, I have been reminded that the point name, library
name and function name have hardcoded limits, and it is now possible
to have them tested by SQL.
Attached is a patch to do so. Thoughts?
--
Michael
Fr
On Sun, Nov 09, 2025 at 11:51:51PM +0100, Mihail Nikalayeu wrote:
> Yes, possible. But the idea is to avoid weird two INJECTION_POINT in a
> row...
FWIW, I think that it's OK to use as style. The backend-side
injection point implementation is currently quite simple, meaning that
it less prone to
On Sun, Nov 09, 2025 at 08:35:55AM +0900, Michael Paquier wrote:
> Yeah. What you are doing would be enough on simplicity ground. The
> test added is also fine enough, it's safe to run even under an
> installcheck. So LGTM to use a minimal implementation.
The patch had a one problem other than
On Fri, 7 Nov 2025 at 07:26, Peter Eisentraut wrote:
>
> On 06.11.25 13:17, Heikki Linnakangas wrote:
> >> @@ -476,7 +476,7 @@ CatCachePrintStats(int code, Datum arg)
> >>
> >> if (cache->cc_ntup == 0 && cache->cc_searches == 0)
> >> continue;/* don't print unused
On Wed, Oct 1, 2025 at 2:16 PM jian he wrote:
>
> hi.
> v19 attached, same as v18.
> repost it so that CFbot can pick up the latest patchset.
hi.
new patch attached, rebase only.
From 97e63d2b7de1fef820305b279d9e5602c82dab53 Mon Sep 17 00:00:00 2001
From: jian he
Date: Mon, 10 Nov 2025 08:39:36
Hi,
As described in the manual[1], PREPARE ... AS command supports MERGE and VALUES
statements, but current psql's tab completion does not suggest them.
I created a patch that enables MERGE and VALUES to appear as tab completion
suggestions.
What do you think?
[1] https://www.postgresql.org/do
On Mon, Nov 10, 2025 at 08:30:47AM +1100, Peter Smith wrote:
> On Sun, Nov 9, 2025 at 1:50 AM Bruce Momjian wrote:
> >
> > On Tue, Nov 4, 2025 at 02:35:29PM +1100, Peter Smith wrote:
> > > On Thu, Jul 24, 2025 at 2:18 AM Bruce Momjian wrote:
> > > ...
> > > > You are right that we are inconsiste
Hello!
нд, 9 ліс 2025, 23:34 карыстальнік Michael Paquier
напісаў:
> It is already possible to achieve that, isn't it? One can define
> side-by-side twice INJECTION_POINT(), then assign a different callback
> to each with a strict control of the execution order.
>
Yes, possible. But the idea i
On Sun, Nov 09, 2025 at 01:53:00PM +0100, Mihail Nikalayeu wrote:
> What if we extend that limitation to allow registration of different
> types for the same injection point? Like 'wait' and 'notice' (tol be
> executed in order they registered).
> It feels like a pretty simple solution and covers m
On Fri, Sep 12, 2025 at 9:46 AM Andres Freund wrote:
+ * It's possible that another backend starts IO on the buffer between this
+ * check and the ReadBuffersCanStartIO(nowait = false) below. In that case
+ * we will synchronously wait for the IO below, but the window for that is
+
On Wed, Nov 5, 2025 at 2:12 AM Antonin Houska wrote:
> Robert Treat wrote:
> > On Tue, Nov 4, 2025 at 9:48 PM jian he wrote:
>
> > > what the expectation of
> > > pg_repackdb --index=index_name, the doc is not very helpful.
> > >
> > > pg_repackdb --analyze --index=zz --verbose
> > > pg_repackdb
On Sun, Nov 9, 2025 at 1:50 AM Bruce Momjian wrote:
>
> On Tue, Nov 4, 2025 at 02:35:29PM +1100, Peter Smith wrote:
> > On Thu, Jul 24, 2025 at 2:18 AM Bruce Momjian wrote:
> > ...
> > > You are right that we are inconsistent. I think is the
> > > accepted way to reference table names, and fo
On Thu, Sep 11, 2025 at 5:46 PM Andres Freund wrote:
> The problem leading to that is that we are currently synchronously waiting for
> IO on a buffer when AsyncReadBuffers() encounters a buffer already undergoing
> IO. If a block is read twice, that means we won't actually have enough IOs in
> fl
Tomas Vondra writes:
> If I set join_collapse_limit=1, then standard_join_search() only sees
> problems of size 2, i.e. (list_length(initial_rels) == 2). And we only
> look at has_join_restriction() *inside* these small problems, i.e. the
> jointree must not be deconstructed in a way that would vi
On 11/8/25 21:36, Tom Lane wrote:
> [ Don't have time to read the v4 patch right now, but a couple
> of quick responses: ]
>
> Tomas Vondra writes:
>> On 9/23/25 21:46, Tom Lane wrote:
>>> I'd be slightly inclined to put the GUC test there, too:
>>>
>>> + if (enable_starjoin_join_search)
>>>
Philip Warner writes:
> On 2025-11-09 04:06, Tom Lane wrote:
>> IIRC you'd need to use --create to prod pg_dump to produce
>> a CREATE DATABASE command as well as any database-level
>> ALTER commands.
> Is that a good approach?
Well, it's what pg_dump has always done, at least since it inherited
Hello, Michael!
Currently injection_point.c says [0]:
> /*
> * Allocate and register a new injection point. A new point should not
> * exist. For testing purposes this should be fine.
> */
What if we extend that limitation to allow registration of different
types for the same injection point?
Hello!
Rebased, excluded committed part, updated revering part.
Best regards,
Mikhail.
From 7ae8d431ec01c1c89a8de675dd05d036e37f714c Mon Sep 17 00:00:00 2001
From: nkey
Date: Thu, 20 Feb 2025 14:50:58 +0300
Subject: [PATCH v12 2/5] Modify the infer_arbiter_indexes function to consider
both indi
Hi everybody,
Apologies for jumping into this conversation.
Our customers have also encountered a similar issue with the
concurrent drop of a dependent object. In our code (in the Greengage
DB), we implemented a fix analogous to one of the first versions of
Bertrand's approach, using locking wit
39 matches
Mail list logo
