On 05.12.25 15:46, Jelte Fennema-Nio wrote:
Calling copyObject fails in C++ with an error like in most setups:
error: use of undeclared identifier 'typeof'; did you mean 'typeid'
This is due to the C compiler supporting used to compile postgres
supporting typeof, but that function actually not
Hi,
On Fri, Dec 05, 2025 at 03:52:40PM -0500, Tom Lane wrote:
> I wrote:
> > Ah. Your script didn't notice the need for const'ification of
> > additional variables in map_locale() :-(.
Ah, right, thanks for letting me know. It would need more work to catch those.
That said I improved the script
On Fri, Dec 5, 2025 at 4:56 PM Masahiko Sawada wrote:
>
> On Thu, Dec 4, 2025 at 11:13 PM Masahiko Sawada wrote:
> >
> > On Thu, Dec 4, 2025 at 7:59 PM shveta malik wrote:
> > >
> > > On Fri, Dec 5, 2025 at 12:23 AM Masahiko Sawada
> > > wrote:
> > > >
> > > > On Thu, Dec 4, 2025 at 1:30 AM sh
Hi Zhijie,
Thanks for the patch. The change looks reasonable. ShutDownSlotSync() has set
SlotSyncCtx->stopSignaled, and SIGUSR1’s procsignal_sigusr1_handler will do
SetLatch(MyLatch) that will in turn wake up ProcessSlotSyncInterrupts(), then
hit if (SlotSyncCtx->stopSignaled) and the slotsync
Mihail Nikalayeu wrote:
> On Thu, Dec 4, 2025 at 6:43 PM Antonin Houska wrote:
> > v26 attached here. It's been rebased and reflects most of the feedback.
>
> Some comments on 0001-0002:
> 1)
>
> > cluster_rel(stmt->command, rel, indexOid, params);
> cluster_rel closes relation, and after it i
On Fri, Dec 5, 2025 at 4:10 AM Amit Kapila wrote:
>
> On Thu, Dec 4, 2025 at 12:12 PM Zhijie Hou (Fujitsu)
> wrote:
> >
> > On Thursday, December 4, 2025 1:58 PM Amit Kapila
> > wrote:
> > >
> > > On Thu, Dec 4, 2025 at 2:04 AM Masahiko Sawada
> > > wrote:
> > > >
> > > > On Tue, Dec 2, 2025 a
Thank you for the detailed feedback.
> > Your tests are using a busy loop with usleep which nowadays is
> > considered as bad practice. There 3 of such places, and I think the
> > first two of them
> > can be replaced with injection point wait.
>
> (This fell off my radar, apologies about that.)
Hi all,
Based on the ongoing discussions on Checkpointer Write Combining [1] and
the last patch v11 discussed in forum, I have re-implemented the logic for
CleanVictimBuffer() and BufferNeedsWALFlush() with a few changes:
static void
CleanVictimBuffer(BufferDesc *bufdesc, bool from_ring, IOContex
Hi All,
test_decoding output plugin skips sending stream_start and stream_end
if no change from a segment of a streamed transaction is sent
downstream. If no change has been sent downstream across all the
streamed segments it doesn't send commit or abort for that streamed
transaction. But I don't s
> I summarized the number of comparisons needed for different 'k':
>
> k = 2, heap: 1, loser tree: 1
> k = 3, heap: 2, loser tree: [1, 2]
> k = 4, heap: [2, 3], loser tree: 2
> k = 5, heap: [2, 4], loser tree: [2, 3]
>
> So if k < 5, the loser tree is never worse than the heap for any input data.
Thanks Tushar for the testing and reports.
On Tue, 2 Dec 2025 at 18:45, tushar wrote:
>
>
>
> On Mon, Dec 1, 2025 at 10:47 PM tushar wrote:
>>>
>>>
>>
>> I have observed that when combining the --globals-only option with certain
>> other switches during a pg_restore - operation fails silently.
On Fri, Dec 05, 2025 at 07:27:50PM -0600, Sami Imseih wrote:
> Attached is the new test module that replaces the custom statistics
> tests currently in the injection points tests. Under test_custom_stats, there
> are two separate modules: one for variable-amount stats and one for
> fixed-amount sta
> On Dec 8, 2025, at 14:26, John Naylor wrote:
>
> Also notice that by attaching a .diff, the CF bot tries and fails to
> apply that to master, and has been complaining that my patch needs a
> rebase. Please don't do that again.
>
Fair point. I wasn’t aware you have created a CF entry.
Best
On Mon, Dec 8, 2025 at 9:52 AM Chao Li wrote:
> First, I changed my direction and implemented the in-place switching in the
> other way, where I did a way like chained-switching. Say starting from item0,
> for example, switching item0 to item5, then check where item5 should be
> switched to, an
On Fri, Dec 5, 2025 at 10:40 PM Nathan Bossart wrote:
> I don't think the proposed improvements are relevant for either of the
> machines you used for your benchmarks. For x86, we've optimized our
> popcount code to use SSE4.2 or AVX-512, and for AArch64, we've optimized it
> to use Neon or SVE.
Hi Shlok. Some review comments for v29-0001 (ALTER PUBLICATION RESET)
==
doc/src/sgml/ref/alter_publication.sgml
Description:
1.
The command ALTER PUBLICATION can change the attributes
- of a publication.
-
+ of a publication. There are several subforms described below.
IMO, y
On Thu, Dec 4, 2025 at 12:50 PM Michael Paquier wrote:
>
> On Wed, Dec 03, 2025 at 09:11:03AM -0500, Greg Burd wrote:
> > On Dec 2 2025, at 11:16 pm, Michael Paquier wrote:
> > I've looked in the email archives a bit and didn't come up with much
> > related to catalog upgrades. I don't know if t
On Thu, Dec 4, 2025 at 4:20 PM Amit Kapila wrote:
>
> On Thu, Dec 4, 2025 at 10:49 AM Dilip Kumar wrote:
> >
> > > ---
> > > I think the conflict history table should not be transferred to the
> > > new cluster when pg_upgrade since the table definition could be
> > > different across major versi
On Sat, Dec 6, 2025 at 11:23 PM Naga Appani wrote:
>
> Hi Ashutosh,
>
> Thanks for the review!
>
> I agree - comparing the exposed members_size against the documented
> thresholds is sufficient for monitoring purposes.
>
> This aligns with the approach taken in v11: exposing the current usage in
>
Hi,
Thanks for reviewing and sharing your feedback.
> 1) Please do not top-post on these lists
Apologies for that. I will make sure to follow this going forward.
> 2) I did not get the exact reason this thread is different from [0]
The purpose of this thread was to share the actual patch address
On Mon, Dec 8, 2025 at 7:34 AM Zhijie Hou (Fujitsu)
wrote:
>
> Hi,
>
> Previously, the slotsync worker used SIGINT to receive a graceful shutdown
> signal from the startup process on promotion. However, SIGINT is also used by
> the LOCK_TIMEOUT handler to trigger a query-cancel interrupt. Given th
On Mon, Dec 8, 2025 at 9:12 AM vignesh C wrote:
>
> On Sat, 6 Dec 2025 at 20:36, Dilip Kumar wrote:
> >
> > On Fri, Dec 5, 2025 at 10:39 AM Dilip Kumar wrote:
> > >
> > > On Thu, Dec 4, 2025 at 8:05 PM vignesh C wrote:
> > > >
> > > > On Wed, 3 Dec 2025 at 16:57, Dilip Kumar wrote:
> > > > >
>
On Sat, 6 Dec 2025 at 20:36, Dilip Kumar wrote:
>
> On Fri, Dec 5, 2025 at 10:39 AM Dilip Kumar wrote:
> >
> > On Thu, Dec 4, 2025 at 8:05 PM vignesh C wrote:
> > >
> > > On Wed, 3 Dec 2025 at 16:57, Dilip Kumar wrote:
> > > >
> > > > On Wed, Dec 3, 2025 at 9:49 AM shveta malik
> > > > wrote:
On Thu, Dec 4, 2025 at 1:11 PM cca5507 wrote:
> I summarized the number of comparisons needed for different 'k':
>
> k = 2, heap: 1, loser tree: 1
> k = 3, heap: 2, loser tree: [1, 2]
> k = 4, heap: [2, 3], loser tree: 2
> k = 5, heap: [2, 4], loser tree: [2, 3]
>
> So if k < 5, the loser tree is
On Thu, Dec 4, 2025 at 1:56 PM vignesh C wrote:
>
> On Thu, 4 Dec 2025 at 07:47, Peter Smith wrote:
> >
> > On Wed, Dec 3, 2025 at 4:47 PM tianbing wrote:
> > >
> > > Hi， Peter，
> > > I have reviewed the v21 patch and noticed that there seems to be a memory
> > > leak.
> > >
> > > +static bool
Hi John,
I played the radix sort again during the weekend.
First, I changed my direction and implemented the in-place switching in the
other way, where I did a way like chained-switching. Say starting from item0,
for example, switching item0 to item5, then check where item5 should be
switched
Hi hackers,
I encountered test_aio failures on RHEL 9 / Rocky Linux 9 systems where
io_uring is disabled by default (io_uring_disabled=2). This is increasingly
common in enterprise environments.
Background:
Red Hat disabled io_uring in 2021 and maintains this policy in RHEL 9
for security reasons
Hi,
> Using "However" here indicates some exception to what's just been
> said, but there is no longer an exception. To write about what the
> 1024 is for, we might need to reverse engineer what that's for. I
> assume it's something like "Clamp at 1024 elements to avoid excessive
> reallocs of the
Hi,
Previously, the slotsync worker used SIGINT to receive a graceful shutdown
signal from the startup process on promotion. However, SIGINT is also used by
the LOCK_TIMEOUT handler to trigger a query-cancel interrupt. Given that the
slotsync worker can access and lock catalog tables while parsing
On Friday, December 5, 2025 8:10 PM Amit Kapila wrote:
>
>
> Fair enough. The patch looks mostly good to me, attached are minor comment
> improvements atop the HEAD patch. I'll do some more testing before push.
>
Thanks for the diff, it looks good to me.
Here are the updated patches for HEAD
On 2025-11-28 18:22, Rahila Syed wrote:
Hi,
I'm attaching the updated patches, which primarily include cleanup and
have been rebased
following the CFbot report.
Thanks for updating the patch!
I observed an assertion failure when forcing a timeout as follows:
```
$ psql
(pid:38587)=#
$ ki
On Fri, Dec 05, 2025 at 12:30:38AM -0500, Corey Huinker wrote:
> Attached is a follow-on to Michael's most recent uncommitted patch,
> changing the errors that I see as "impossible" to elogs. However, I agree
> that they don't add significant workload to the translations, and most
> input functions
> On Dec 8, 2025, at 08:28, David Rowley wrote:
>
> + * in grow_memtuples(). However, we don't consider array sizes
> + * less than 1024.
>
> Using "However" here indicates some exception to what's just been
> said, but there is no longer an exception. To write about what the
> 1024 is for,
On Mon, 8 Dec 2025 at 12:10, David G. Johnston
wrote:
>
> On Sun, Dec 7, 2025 at 3:09 PM David Rowley wrote:
>>
>> The comment is effectively
>> explaining that we don't want to make the array big enough so that a
>> malloc will always be required.
>
>
> Doesn't what you are saying contradict bot
The following review has been posted through the commitfest application:
make installcheck-world: not tested
Implements feature: not tested
Spec compliant: not tested
Documentation:not tested
Hi,
I tested the patch on top of current master. It applies cleanly and
buil
On Sun, Dec 07, 2025 at 10:07:45PM +0100, Alvaro Herrera wrote:
> Hmm, as I recall it's quite intentional that the index on a partitioned
> table is marked !indisvalid; such indexes are only supposed to be marked
> valid once indexes on all partitions have been attached. As I recall,
> if you remo
On Sun, Dec 7, 2025 at 3:09 PM David Rowley wrote:
> The comment is effectively
> explaining that we don't want to make the array big enough so that a
> malloc will always be required.
Doesn't what you are saying contradict both the formula (the +1
post-division) and the comments:
/*
* Initial
On Sat, Dec 6, 2025 at 9:44 PM Peter Geoghegan wrote:
> Since this ignore_killed_tuples change is also very simple, and also
> seems like an easy win, I think that it can be committed as part of
> the second patch. Without it needing to wait for too much more
> performance validation.
My plan is
Hello!
On Sun, Dec 7, 2025 at 10:07 PM Álvaro Herrera wrote:
> I'd rather consider the idea of avoiding indexes marked !indisvalid on
> partitioned tables as arbiter lists ... but then we need to verify the
> scenario where there is one, and INSERT ON CONFLICT runs concurrently
> with ALTER INDEX
On Sun, 7 Dec 2025 at 21:34, cca5507 wrote:
> I find that the initial size of memtuples array must be more than
> ALLOCSET_SEPARATE_THRESHOLD
> is not for lower overhead of allocation, but for a bug fixed in
> 8ea3e7a75c0d22c41c57f59c8b367059b97d0b66.
>
> Attach a new patch.
I find the current
On 2025-Dec-06, Donghang Lin wrote:
> The issue itself in fact doesn't depend on BtreeCheckState's snapshot
> member. So the fix for 14-16 can make snapshot a local variable in
> the bt_check_every_level function to make the scope small. Do you
> think it's worth changing it to a local variable?
On 2025-Dec-06, Mihail Nikalayeu wrote:
> Such commands add an indisready, but not indisvalid index on pt, which
> is added to to the list of potential arbiters.
> It happens because of [0].
>
> From my perspective, the correct solution - is to just remove the
> error message, because it looks ob
On Sat, 6 Dec 2025 at 09:12, Cary Huang wrote:
> > within each worker process.
> >
> >
> > +
> > +
> > +In a parallel tid range scan, the range of
> > blocks
> > +will be subdivided into smaller ranges which are shared among the
> > +cooper
Jelte Fennema-Nio writes:
> Calling copyObject fails in C++ with an error like in most setups:
> error: use of undeclared identifier 'typeof'; did you mean 'typeid'
> This is due to the C compiler supporting used to compile postgres
> supporting typeof, but that function actually not being present
Thank you for your kind review!
Before I continue with the implementation, I would like to address
your concerns and discuss it further and see if it's worth carrying
on.
1) Will do!
2) The issue is the following:
```
./../../src/test/regress/pg_regress --temp-instance=./tmp_check
--inputdir=. -
On 05.12.2025 16:33, Peter Eisentraut wrote:
> There are many PG_GETARG_* calls, mostly around gin, gist, spgist code,
> that are commented out, presumably to indicate that the argument is
> unused and to indicate that it wasn't forgotten or miscounted. Example:
>
> ...
> StrategyNumber strat
I took a look at the earlier comments.
> +/* Read and verify the hash key */
> +if (!pgstat_read_chunk(fd, (void *) key, sizeof(PgStat_HashKey)))
> +return;
> [...]
> +/* Write the hash key to identify this entry */
> +pgstat_write_chunk(fd, (void *) key, sizeof(PgStat_Hash
Chao Li writes:
> On Dec 7, 2025, at 03:00, Tom Lane wrote:
>> I've been thinking of that as an independent issue. But what I'm
>> inclined to do is add a symbol to date.h, say like
>> ...
> That’s even better than resolving my comment.
Cool, done like that then.
regar
Hello, comments so far on 0004:
---
> ind_oids_new = build_new_indexes(NewHeap, OldHeap, ind_oids_old);
I think the biggest issue we have so far -
repack_decode_concurrent_changes is not called while new indexes are
built (the build itself creates a huge amount of WAL and takes days
sometimes). L
On Tue Feb 11, 2025 at 8:42 PM CET, Andres Freund wrote:
Not sure that's quite the right thing to do for postmaster. What I'd start
with is to increase the soft limit to
"already used files" + max_files_per_process.
I needed to rebase this patch, and that made me finally take the time to
do t
On Mon, Dec 1, 2025 at 2:33 PM Tom Lane wrote:
>
> jian he writes:
> > The attached patch implements the $subject.
>
> Does this cover the case where a BEFORE UPDATE trigger has modified
> columns that were not mentioned in UPDATE...SET?
>
> regards, tom lane
hi.
in ExecI
The following review has been posted through the commitfest application:
make installcheck-world: tested, passed
Implements feature: not tested
Spec compliant: not tested
Documentation:not tested
Hi,
Just reviewed the patch and it seems harmless. I think we should com
Greetings,
My main driver here is to allow the creation of Holdable portals at the
protocol level for drivers. Currently the only way to create a holdable
cursor is at the SQL level.
DECLARE liahona CURSOR WITH HOLD FOR SELECT * FROM films;
The JDBC driver has an option in the API to have resul
On Tue, 18 Feb 2025 at 10:29, Jelte Fennema-Nio wrote:
> > So I think the solution in 0002, while debatable if strictly correct,
> > is the only fix that we can easily do. Also I personally believe the
> > behaviour resulting from 0002 is totally correct: The new behaviour
> > would be that if a t
On Thu, Dec 4, 2025 at 4:35 PM Kirill Reshke wrote:
> On Thu, 4 Dec 2025 at 03:56, Masahiko Sawada
> wrote:
>
> > The proposed patch requires us to create one function per option. I'm
> > concerned that it could cause bloating functions and be overkill just
> > to save changing a separate list.
You can see how much time a subscriber will need to get all files which
were added with this.
Run first time and will create 10 thousand tables, publish them and measure
how much time to get all tables Ready on pg_subscription_rel.
Run again to add more 10 thousand tables and see that time will inc
On Sat, 6 Dec 2025 at 21:40, Tom Lane wrote:
> Yeah. However, selecting different PG_PRINTF_ATTRIBUTE values for
> C and C++ mode seems to get the job done. I've confirmed the attached
> silences these warnings for me when mixing-and-matching gcc and clang.
I've definitely run into this myself
hi.
jsonb_path_query('"abc,def,ghi,jkl"',
'$.split_part(",", 2)')
"ghi"
the return value should be
"def"
?
string .
ltrim(characters)
string
String with the longest string containing only spaces or the
c
On Fri, 17 Jan 2025, 17:22 Vladimir Sitnikov,
wrote:
> Kirill,
>
> cursor does not provide a way to limit the fetch size based on the memory
> consumption.
> Imagine a table like (id int8, value jsonb).
> If we use "fetch 1000", then it might require 1GiB on the client if every
> row contains 1Mi
Hi,
I find that the initial size of memtuples array must be more than
ALLOCSET_SEPARATE_THRESHOLD
is not for lower overhead of allocation, but for a bug fixed in
8ea3e7a75c0d22c41c57f59c8b367059b97d0b66.
Attach a new patch.
--
Regards,
ChangAo Chen
v2-0001-Fix-incorrect-comments-in-tuplesort
60 matches
Mail list logo
