ore calling it. If
that's the case, why don't we just release all LWLocks in shmem_exit
or in AtProcExit_Buffers before calling AbortBufferIO()? I think it's
sufficient that AtProcExit_Buffers calls it at the beginning. (The
comment for the funcgtion needs editing).
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
by the error is unacceptable, but the ENOENT is just
ignorable for the same reason.
As the result, I agree to (a) (fix rounding), and (c) (test
wrap-around before writing WAL) but I'm not sure for others. And
additional fix for ignorable ENOENT is needed.
What do you think about this?
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
Hello.
On Wed, Jul 24, 2019 at 2:31 AM Alvaro Herrera wrote:
>
> On 2019-Jul-23, Tom Lane wrote:
>
> > Kyotaro Horiguchi writes:
>
> > > My investigation convinced me that there is no way for a process
> > > to detect wheter it is running as a service (exce
I found that CF-bot complaining on this.
Seems that some comment fixes by the recent 21039555cd are the
cause.
No substantial change have been made by this rebasing.
regards.
On Fri, Jul 12, 2019 at 5:37 PM Kyotaro Horiguchi
wrote:
>
> At Fri, 12 Jul 2019 17:30:41 +0900 (Tokyo Standar
ng direction is different from index direction, we must
+ * skip right away, but _bt_skip requires a starting point.
+ */
+if (direction * indexonlyscan->indexorderdir < 0 &&
+ !node->ioss_FirstTupleEmitted)
I'm confused by this. "direction" there is the physical scan
direction (fwd/bwd) of index scan, which is already compensated
by indexorderdir. Thus the condition means we do that when
logical ordering (ASC/DESC) is DESC. (Though I'm not sure what
"index direction" exactly means...)
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
Sorry, there's a too-hard-to-read part.
At Thu, 25 Jul 2019 20:17:37 +0900 (Tokyo Standard Time), Kyotaro Horiguchi
wrote in
<20190725.201737.192223037.horikyota@gmail.com>
> Hello.
>
> At Wed, 24 Jul 2019 22:49:32 +0200, Dmitry Dolgov <9erthali...@gmail.com>
ng knobs.
Documentation (psql-ref.sgml) has another place that needs the
same amendment.
Looking the output for -t, -0, -A or something like, we might need
to introduce result-set separator.
# -eH looks broken for me but it would be another issue.
Valid setting of FETCH_COUNT disables this feature. I think it is
unwanted behavior.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
sql.org/message-id/alpine.DEB.2.21.1904132231510.8961@lancre
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
>From 6dbc907142d7aaba7f02e6585e3fa407511725bc Mon Sep 17 00:00:00 2001
From: Kyotaro Horiguchi
Date: Fri, 26 Jul 2019 12:51:56 +0900
Subject: [PATCH] Don't parse furth
ng..
The function returns NULL for NULL input (STRICT behavior) but
returns (NULL, NULL) for undefined timeline. I don't think the
differene is meaningful.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
-
0/16ff | 00020016 |16777214
0/1700 | 00020016 |16777215
0/1701 | 00020017 | 0
regards.
[1]: https://www.postgresql.org/message-id/20190725193808.1648ddc8@firost
--
Kyotaro Horiguchi
NTT Open Source Softw
nd probably down to every version of
> postgres
> since the feature was implemented...
>
> I think that fixing this should be a separate bug report and
> patch. I'll try to look at it.
Ah, I didin't notieced that. Thanks for the explanation.
> Thanks for the feedback. Attached v3 with further documentation
> updates.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
*/
+if (last && pset.timing)
I'm not sure whether we reached any consensus with ths
behavior. This means the timing includes result-printing time of
other than the last one. If we don't want include printing time
at all, we can exclude it with a small amount of additional
complexity.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
Hello.
On 2019/07/29 4:17, Dmitry Dolgov wrote:>> On Thu, Jul 25, 2019 at 1:21 PM
Kyotaro Horiguchi wrote:
> Yeah, will change both (hopefully soon)
Thanks.
>> + /*
>> + * XXX: In case of index scan quals evaluation happens after
>> + *
e a definitive status
> * it seems related, but maybe the "wal_status" should be set as "lost"
> only when the slot has been deactivate ?
Agreed. While replication is active, if required segments seems
to be lost once, delayed walreceiver ack can advance restart_lsn
to "safe" zone l
At Tue, 30 Jul 2019 21:30:45 +0900 (Tokyo Standard Time), Kyotaro Horiguchi
wrote in
<20190730.213045.221405075.horikyota@gmail.com>
> I attach the revised patch. I'll repost the polished version
> sooner.
(Mainly TAP test and documentation, code comments will be
At Tue, 30 Jul 2019 21:30:45 +0900 (Tokyo Standard Time), Kyotaro Horiguchi
wrote in
<20190730.213045.221405075.horikyota@gmail.com>
> I attach the revised patch. I'll repost the polished version
> sooner.
This is the revised patch.
- Status criteria has been chang
with
ext-stats names, but it's the place for target value.
+++ b/src/include/nodes/nodes.h
T_CallStmt,
+ T_AlterStatsStmt,
I think it should be immediately below T_CreateStatsStmt.
+++ b/src/include/nodes/parsenodes.h
+ boolmissing_ok;/* skip error if statistics object is missing */
Should be very trivial, but many bool members especially
missing_ok have a comment having "?" at the end.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
Hello.
At Fri, 2 Aug 2019 11:35:06 +1200, Thomas Munro wrote
in
> On Sat, Jul 27, 2019 at 6:26 PM Noah Misch wrote:
> > On Thu, Jul 25, 2019 at 10:39:36AM +0900, Kyotaro Horiguchi wrote:
> > > No substantial change have been made by this rebasing.
> >
> > Thank
spondence between
top-level query and queryId (in nested or multistatement queries)
in this patch. pg_stat_statements will allow us to do the same
thing by having additional uint64[MaxBackends] array in
pgssSharedState, instead of expanding PgBackendStatus array in
core by the same size.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
cord
> length at B8D/44FFF740: wanted 24, got 0
The shutdown record is written at the end of crash recovery of
the master, then replicated. As the result the subsequent bytes
are inconsistent with the record.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
l ...)
-> Limit (cost=0.20..0.40 rows=10 width=4) (actual ...)
-> Seq Scan on t0 (cost=0.00..2.00 rows=100 width=4) (actual ...)
That's it for now.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
inder 1),
GAP (modulus 4, remainder 2)
Or
Partitions: c1 FOR VALUES WITH (modulus 4, remainder 0),
c3 FOR VALUES WITH (modulus 4, remainder 3),
Gaps: (modulus 4, remainder 1), (modulus 4, remainder 2)
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
x27;m not sure, but it's not likely that someone adds it without
actually stumbling on space-containing paths with the ealier
version. Anyway if we shouldn't touch this unless the existing
code makes actual problem.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
versions of VC might need it. I confirmed
that the extra " actually harms at least VC2019 and the code
removal in the patch works. As for the replace comment, I'm not
sure it is needed since I think quoting is not the task for
AddLibrary/AddIncludeDir in the first place (and AddIncludeDir
doesn't have the same comment).
Now I'm trying to install VS2015 into my alomost-filled-up disk..
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
Hello.
At Thu, 08 Aug 2019 12:15:38 +0900 (Tokyo Standard Time), Kyotaro Horiguchi
wrote in
<20190808.121538.87367461.horikyota@gmail.com>
> At Wed, 7 Aug 2019 12:14:48 +0300, Dmitry Igrishin wrote
> in
> > > -if ($lib =~ m/\s/)
> > > -{
&g
sonable. By the way, the patch doesn't touch
get_relation_constraints(), but I suppose it can use the modified
partition constraint qual already stored in rel->partition_qual
in set_relation_partition_info. And we could move constifying to
set_rlation_partition_info?
Also, I'd like to see comments that the partition_quals is
already varnode-fixed.
And 0002, yeah, just +1 from me.
[1]
https://www.postgresql.org/message-id/20190409.173725.31175835.horiguchi.kyot...@lab.ntt.co.jp
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
At Fri, 9 Aug 2019 14:02:36 +0900, Amit Langote wrote
in
> On Fri, Aug 9, 2019 at 1:17 PM Amit Langote wrote:
> > On Fri, Aug 9, 2019 at 12:09 PM Kyotaro Horiguchi
> > wrote:
> > > At Thu, 8 Aug 2019 14:50:54 +0900, Amit Langote wrote:
> > > > When workin
ice but '|' make my eyes twitch (a bit). Couldn't we
use other delimites like (), ##, or // ? (I like {} for use in
this patch.)
Any opinions?
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
ess/transam/xact.c:
BlockStateAsString(TBlockState blockState)
I foundnd them by
find $(TOP) -type f -exec egrep -nH -e '^(static )?const char \*' {} +
then eyeballing on the first ones. I don't know an automated way
to detect such possibly-loose constness of variables or
functi
em, but development has shown
> that it could also be used for other CachedPlans, as long as it is a
> saved plan.
> > - Is this a WIP patch or the final patch? Because I can see TODO and
> > non-standard
> > comments in the patch.
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
ing-syncs are useless on wal-replay, but that
is needed for commit-prepared.
> On Thu, Jul 25, 2019 at 10:39:36AM +0900, Kyotaro Horiguchi wrote:
> > --- a/src/backend/access/heap/heapam_handler.c
> > +++ b/src/backend/access/heap/heapam_handler.c
> > @@ -715,12 +702,6 @@ he
Hi,
At Wed, 7 Aug 2019 10:20:09 +0300, Surafel Temesgen
wrote in
> Hi
> On Wed, Aug 7, 2019 at 6:11 AM Kyotaro Horiguchi
> wrote:
>
> >
> > I have some comments.
> >
> > This patch uses distinct parameters for exact number and
> > percentage
Hello.
At Mon, 19 Aug 2019 18:59:59 +0900 (Tokyo Standard Time), Kyotaro Horiguchi
wrote in
<20190819.185959.118543656.horikyota@gmail.com>
> > The comment material being deleted is still correct, so don't delete it.
> > Moreover, the code managing rd_firstRelfile
Hello. New version is attached.
At Mon, 19 Aug 2019 18:59:59 +0900 (Tokyo Standard Time), Kyotaro Horiguchi
wrote in
<20190819.185959.118543656.horikyota@gmail.com>
> Thank you for taking time.
>
> At Sat, 17 Aug 2019 20:52:30 -0700, Noah Misch wrote in
> <
Thank you for the suggestion, Heikki.
At Mon, 29 Jul 2019 22:39:57 +0300, Heikki Linnakangas wrote
in
> On 12/07/2019 10:10, Kyotaro Horiguchi wrote:
> >> Just FYI, to me this doesn't clearly enough look like an improvement,
> >> for a change of this size.
> >
from it.
As far as looking the attached spitest.c, it seems that the
overhead comes from cursor operation, not from SPI. As far as
spitest.c goes, cursor is useless. "SQL" and C/coreAPI seem to
be scanning over the result from *a single* query. If that's
correct, why don't you use SPI_execute() then scan over
SPI_tuptable?
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
ication. I thougt that we might be
able to change wal_level from scalar to bitmap but it breaks
backward compatibility..
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
ple() would be what
you are seeking for.
>/* ... otherwise see if we have more array keys to deal with */
>} while (so->numArrayKeys && _bt_advance_array_keys(scan, dir));
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
Hello.
At Mon, 19 Aug 2019 23:03:14 -0700, Noah Misch wrote in
<20190820060314.ga3086...@rfd.leadboat.com>
> On Mon, Aug 19, 2019 at 06:59:59PM +0900, Kyotaro Horiguchi wrote:
> > At Sat, 17 Aug 2019 20:52:30 -0700, Noah Misch wrote in
> > <20190818035230.gb3021...@rfd
At Tue, 27 Aug 2019 15:49:32 +0900 (Tokyo Standard Time), Kyotaro Horiguchi
wrote in
<20190827.154932.250364935.horikyota@gmail.com>
> 128GB shared buffers contain 16M buffers. On my
> perhaps-Windows-Vista-era box, such loop takes 15ms. (Since it
> has only 6GB, the test
Hello.
At Sun, 25 Aug 2019 22:08:43 -0700, Noah Misch wrote in
<20190826050843.gb3153...@rfd.leadboat.com>
noah> On Thu, Aug 22, 2019 at 09:06:06PM +0900, Kyotaro Horiguchi wrote:
noah> > At Mon, 19 Aug 2019 23:03:14 -0700, Noah Misch
wrote in <20190820060314.ga3086..
Hello, Noah.
At Tue, 27 Aug 2019 15:49:32 +0900 (Tokyo Standard Time), Kyotaro Horiguchi
wrote in
<20190827.154932.250364935.horikyota@gmail.com>
> I'm not sure whether the knob shows apparent performance gain and
> whether we can offer the criteria to identify the proper
commit,
commit_delay (or synchronous_commit) works that way.
> Can this idea benefit from WAL writing performance?
> If it's OK to improve, I want to do modification.
> How do you think of it?
So the proposal seems to be already achieved. If not, could you
elaborate the pro
> > Peter's review comments due to other tasks. Does anyone think it's worth
> > resuming this?
> >
> > https://www.postgresql.org/message-id/7E37040CF3804EA5B018D7A022822984@maumau
>
> And I was looking for this thread a couple of lines ago :)
> Thanks.
# Is there any means to view the whole of a thread from archive?
# I'm a kind of reluctant to wander among messages like a rat in
# a maze:p
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
At Tue, 3 Sep 2019 02:42:19 +, "Tsunakawa, Takayuki"
wrote in
<0A3221C70F24FB45833433255569204D1FD1609C@G01JPEXMBYT05>
> From: Kyotaro Horiguchi [mailto:horikyota@gmail.com]
> > Since we are allowing OPs to use arbitrary command as
> > archive_command, p
:
The changes of documentation are seen in the following html files.
doc/src/sgml/html/warm-standby.html#STREAMING-REPLICATION-SLOTS
doc/src/sgml/html/runtime-config-replication.html#GUC-MAX-SLOT-WAL-KEEP-SIZE
doc/src/sgml/html/view-pg-replication-slots.html
One annoyance
have
such feature, it would be something like "psql server" or
"reconnectable session" of frontend protocol. Both seem too much
or leading to something dangerous.
> > Or just
> > write a bespoke testing tool.
> >
> >
>
> ... that's pretty much where we came in.
Agreed. And we can add anything PostgreSQL or test specific
features to this.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
ion didn't conflict with the current
master, though) and changed the status to "Needs Review".
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
>From 15e3d095b89e9a5bb8025008d1475107b340cbd4 Mon Sep 17 00:00:00 2001
From: Kyotaro Horiguchi
Date: Tue, 12 Sep 2017 13:
At Mon, 11 Dec 2017 20:07:53 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20171211.200753.191768178.horiguchi.kyot...@lab.ntt.co.jp>
> > The attached PoC patch theoretically has no impact on the normal
> > code paths and just brings gain in async cases.
>
>
t_line or other
similar functions.
At last, just a simple comment.
- point_eq_point() assumes that NaN == NaN. This is an inherited
behavior from old float[48]_cmp_internal() but it's not a
common treat. point_eq_point() needs any comment about the
special definition as float[48]_cmp_internal has.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
gt; status = "streaming";
> else
> status = "keeping";
> else
> status = "may_lost";
> This duplicates an existing active field, but I think it's useful as slot
> status description.
> wal_status str
t is not
measured from slot's restart_lsn, but from checkpoint LSN.
*1: As I mentioned upthread, I think that at least the
"pg_replication_slots.min_keep_lsn" is arguable since it shows
the same value for all slots and I haven't found no other
appropriate place.
> Fwiw I think there's a real need for this feature so I would like to
> get it in for Postgres 11.
It encourages me a lot. Thanks.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
gt; I think that there might be a much simpler solution to this, which
> > is to just remove make_inh_translation_list's tests of attinhcount,
> > as per attached.
>
> I went ahead and pushed that, with an isolation test based on the
> one Horiguchi-san submitted but covering
my original setting is not so problematic so I
reverted it.
> Not sure, but that might be another situation in which "patch"
> works and "git apply" doesn't. (Feeling too lazy to test it...)
I was also afraid of that as I wrote upthread but it seems also a
needless fear.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
t it the best way to improve this. I am trying to avoid
> dealing with EPSILON related issues in the scope of these changes.
Right. And I don't have a good idea here..
> > At last, just a simple comment.
> >
> > - point_eq_point() assumes that NaN == NaN. This is an inherited
required but we cannot predict that until
actually acquire a tuple during execution.
Please find the attached patch.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
*** a/src/backend/optimizer/path/indxpath.c
--- b/src/backend/optimizer/path/indxpath.c
***
***
At Thu, 18 Jan 2018 12:57:38 +0500, Andrey Borodin wrote
in <62c2b9a0-51bc-40ff-9bca-f203784cb...@yandex-team.ru>
> Hello!
> > 18 янв. 2018 г., в 10:48, Kyotaro HORIGUCHI
> > написал(а):
> >
> > Gist imposes the ninth strategy to perform index only scan but
&g
the problem, you surely should add some (haven't
> checked the patch in detail, sorry ;p ).
Uggg.. I'm beaten again.. You're definitely right!
It was a bit hard to find the way to cause the failure without
extension but the first attached file is that.
regards,
--
Kyotaro Horig
line_parallel still requires some kind
of tolerance to work properly. Since this patch is an
imoprovement patch, I think we can change it to the vector method.
The problem of line_distance is an existing one so we can leave
it alone. It returns 0.0 for the most cases but it is a
long-standin
Hello.
At Fri, 19 Jan 2018 01:16:56 +0300, Alexander Korotkov
wrote in
> On Thu, Jan 18, 2018 at 8:48 AM, Kyotaro HORIGUCHI <
> horiguchi.kyot...@lab.ntt.co.jp> wrote:
> > Index only scan is not usable in the case since the first index
> > column cannot be rechec
ot data should updated with that instead of the
> > flush LSN).
>
> -0.5, I think this just adds complexity instead of fixing the underlying
> problem.
On the other hand if one logical record must be read from single
source, we need any means to deterrent wal-recycling on the
primary side. Conducting that within the primary side is rejected
by consensus. (There might be smarter way to do that, though.) To
do that from the standby side, just retarding write feedbacks or
this kind of special feedback would be needed. In any way it is
necessary to introduce WAL-record awareness into WAL shipping
process and it's seemingly large complexity.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
Anyway, mails via -hackers ML doesn't suffer so, what Amit (and
I) side can do by myself is one of the following.
- Being careful to reply to the mails comming via the ML.
- Remove the added header by hand..
And I'd like to ask David to check out his mail environment so
that SPF record i
r method.
>
> I am making line_parallel() use line_interpt_line() in response to
> your first remark. Vector based algorithm is probably better for
> every use of line_interpt_line(), but it is a bigger change with more
> user visible effects.
I'm fine with that.
> > The problem of line_distance is an existing one so we can leave
> > it alone. It returns 0.0 for the most cases but it is a
> > long-standing behavior.. (Anyway I don't find a reasonable
> > definition of the distance between very-nearly parallel lines..)
>
> Exact comparison with 0.0 instead of FPzero() would also be an
> improvement for line_distance(), but I am not doing it now.
reagrds,
--
Kyotaro Horiguchi
NTT Open Source Software Center
ined, so the existing
behavior is not affected.
- The syntax is existing but just not implemented so I don't
think no arguemnts needed here.
- It undocumentedly inhibits the usage of the namespace prefix
"pgdefnamespace.pgsqlxml.internal" but I believe no one can
notice that.
Hello,
At Fri, 19 Jan 2018 11:28:58 +, Greg Stark wrote in
> On 19 January 2017 at 09:37, Kyotaro HORIGUCHI
> wrote:
> >
> > Though I haven't look closer to how a modification is splitted
> > into WAL records. A tuple cannot be so long. As a simple test, I
>
At Fri, 19 Jan 2018 18:24:56 +0900, Michael Paquier
wrote in <20180119092456.ga1...@paquier.xyz>
> On Fri, Jan 19, 2018 at 10:54:53AM +0900, Kyotaro HORIGUCHI wrote:
> > On the other hand if one logical record must be read from single
> > source, we need any means to deter
they don't check canreturn on attribute basis. (And they don't
try index-only on GiST.)
On the other hand the fix patch applies to all affected versions.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
>From 0e221b3396215d67167622cbc07d04b5185d6f66 Mon Sep 17 00:00
Hello.
At Wed, 24 Jan 2018 10:30:39 +0100, Pavel Stehule
wrote in
> Hi
>
> 2018-01-23 8:13 GMT+01:00 Kyotaro HORIGUCHI > I have three comments on the behavior and one on documentation.
> >
> > 1. Lack of syntax handling.
> >
> > ["'" [^&#x
I am fine with the basic test cases added by the
current patch.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
Thank you for kindly noticing me of that.
At Mon, 29 Jan 2018 11:07:31 +1300, Thomas Munro
wrote in
> On Thu, Jan 11, 2018 at 7:59 PM, Kyotaro HORIGUCHI
> wrote:
> > [new patch set]
>
> FYI this is still broken:
>
> test ddl ... FAILED
>
>
Hello,
At Mon, 29 Jan 2018 19:26:34 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20180129.192634.217484965.horiguchi.kyot...@lab.ntt.co.jp>
> While rechecking the patch, I fixed the message issued on losing
> segments in 0001, revised the TAP test since I found
ot;a in (null)" doesn't match the (null, 0)
row so it donesn't harm finally, I don't think this is a right
behavior. null in an SAOP rightop should be just ignored on
partition pruning. Or is there any purpose of this behavior?
- In extract_bounding_datums, clauseinfo is s
I'm not sure that adjusting the intersection to the
tips of the segment is good or not. Adjusting onto the line
can be better in another case. lseg_interpt_lseg, for
instance, checks lseg_contain_point on the line parameter of
lseg_interpt_line.
# I'll be back later..
regards
--
Kyotaro Horiguchi
NTT Open Source Software Center
At Wed, 31 Jan 2018 13:09:09 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20180131.130909.210233873.horiguchi.kyot...@lab.ntt.co.jp>
> At Sun, 21 Jan 2018 21:59:19 +0100, Emre Hasegeli wrote
> in
> > New versions are attached including all changes we discusse
I happend to see a strange geometric calcualtion on master/HEAD.
CREATE TABLE t (l1 line, l2 line);
INSERT INTO t
(SELECT line(point(0, 0), point(x, y)), line(point(0,0), point(-y, x))
FROM (SELECT random() x, random() y FROM generate_series(0, 1000)) AS a);
SELECT l1?-|l2 AS is_perp, l1, l2
Hello,
At Wed, 31 Jan 2018 17:33:42 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20180131.173342.26333067.horiguchi.kyot...@lab.ntt.co.jp>
> 0003: This patch replaces "double" with float and bare arithmetic
> and comparisons on double to special functions
eing
> where is the bright line between "IN ('today')" and "IN (random())".
>
> regards, tom lane
The patch leaves the ambiguity of values like 'today' but doesn't
accept arbitrary functions. Howerver, it needs ad
orkersToExit(pcxt);
> > RESUME_INTERRUPTS();
>
> Actually the backend is getting stuck in
> choose_next_subplan_for_leader(), in LWLockAcquire(), waiting for the
> hanging worker to release the pstate->pa_lock. I think there is
> nothing wrong in this, because it is assumed that LWLock wait is going
> to be for very short tiime, and because of this bug, the lwlock waits
> forever.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
At Tue, 06 Feb 2018 13:34:19 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20180206.133419.02213593.horiguchi.kyot...@lab.ntt.co.jp>
> At Mon, 5 Feb 2018 15:29:27 +0530, Amit Khandekar
> wrote in
> > On 2 February 2018 at 20:46, Robert Haas wrote:
> > &g
seems to be the case.
>
> How do you suggest we go about fixing this? The test in question is
> important, I've caught actual bugs in the implementation with it,
> because it checks that vacuum effectively frees up space.
>
> I'm thinking this vacuum test could be put on its own parallel group
> perhaps? Since I can't reproduce it, I can't know whether that will
> fix it, but it seems sensible.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
At Tue, 6 Feb 2018 14:50:01 +0900, Masahiko Sawada
wrote in
> On Mon, Dec 11, 2017 at 8:15 PM, Kyotaro HORIGUCHI
> wrote:
> > I considered dshash for pgstat.c and the attached is a *PoC*
> > patch, which is not full-fledged and just working under a not so
> &
At Tue, 6 Feb 2018 11:56:27 +0530, Amit Khandekar
wrote in
> On 6 February 2018 at 10:11, Kyotaro HORIGUCHI
> wrote:
> >> At Mon, 5 Feb 2018 15:29:27 +0530, Amit Khandekar
> >> wrote in
> >> > Attached is a patch that fixes this issue on the above lin
At Tue, 06 Feb 2018 19:24:37 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20180206.192437.229464841.horiguchi.kyot...@lab.ntt.co.jp>
> At Tue, 6 Feb 2018 14:50:01 +0900, Masahiko Sawada
> wrote in
> > The implementation of autovacuum work-item has been
#x27;s a bit uneasy that pa_next_plan can be -1 but it looks
perfect to me.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
Hello,
At Tue, 6 Feb 2018 10:41:22 -0300, Claudio Freire
wrote in
> On Tue, Feb 6, 2018 at 10:18 AM, Claudio Freire
> wrote:
> > On Tue, Feb 6, 2018 at 4:35 AM, Kyotaro HORIGUCHI
> > wrote:
> >>> It's starting to look like a timing effect indeed.
>
ail.com
[2] See attached perl script. I hope it is correct.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
#! /usr/bin/perl
$maxmem=1024 * 4;
#=
print "exponential sized strategy\n";
$ss = 64;
$ts = 0;
$sumiteritem = 0;
for ($i = 1 ; $ts < $maxmem ; $i++) {
$ss
be able to rely on the
existence of DSM at the present since over a year we had no
complain despite the fact that DSM is silently turned on? And
apart from that we are ready to remove 'none' from the options of
dynamic_shared_memory_type right now?
If I may rely on DSM, fallback stuff would not be required.
> regards, tom lane
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
At Thu, 08 Feb 2018 18:04:15 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20180208.180415.112312013.horiguchi.kyot...@lab.ntt.co.jp>
> > > I suggest we remove support for dynamic_shared_memory_type = none first,
> > > and see if we get any complaints.
dynamic_shared_memory_type.
We can reproduce this by letting initdb set
dynamic_shared_memory_type=none regardless of actual
availability. (Second attached) and just "make check".
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
diff --git a/src/backend/optimizer/plan/plann
I don't find documentation about this behavior.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
diff --git a/src/bin/initdb/initdb.c b/src/bin/initdb/initdb.c
index 2efd3b7..1f7f2aa 100644
--- a/src/bin/initdb/initdb.c
+++ b/src/bin/initdb/initdb.c
@@ -915,7 +915,7 @@ test_co
Hello.
At Thu, 08 Feb 2018 18:21:56 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20180208.182156.96551245.horiguchi.kyot...@lab.ntt.co.jp>
> I suppose that the problem has not been resolved yet..
I found several bugs during studying this but my conclusion here
is that the
parallel safety issue.
>
> Committed the patch.
Thank you Robert for Committing, and Peter and Michael for the
comments and supplement.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
Hello,
At Mon, 12 Feb 2018 22:28:15 +0900, Michael Paquier wrote
in <20180212132815.gb18...@paquier.xyz>
> On Fri, Feb 09, 2018 at 05:08:23PM +0900, Kyotaro HORIGUCHI wrote:
> > > postgres: max_wal_senders must be less than max_connections
> >
> > I think t
At Mon, 12 Feb 2018 22:05:36 +0900, Michael Paquier wrote
in <20180212130536.ga18...@paquier.xyz>
> On Fri, Feb 09, 2018 at 05:06:35PM +0900, Kyotaro HORIGUCHI wrote:
> 4 regression tests fail when using dynamic_shared_memory_type=none:
> join, aggregates, select_parallel and
Hello.
I happend to find that the comment on formdesc is missing
pg_subscription. Please find the attached patch (I'm sure:) to
fix that .
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c
lecting default max_connections ... 100
| selecting default shared_buffers ... 128MB
| - selecting dynamic shared memory implementation ... posix
Even though probing can end with failure in the case of [3], it
will not be a problem with the patch [4].
regards.
--
Kyotaro Horiguch
The codes that requires PGSempaphore is inactivated when the
symbol is defined in all other places so it seems to me that we
ought to refrain from using it there, too. The attched patch does
that.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
diff --git a/src/backend/storage
Please ignore this thread and see the following one instead.
https://www.postgresql.org/message-id/20180215.201107.78574525.horiguchi.kyot...@lab.ntt.co.jp
At Thu, 15 Feb 2018 20:07:03 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI
wrote in
<20180215.200703.242189033.horiguchi.k
c and spin.h don't aggree on the necessity of pg_sema.h
when HAVE_SPINLOCKS is defined. I post a mail about this issue
separately.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
remove_useless_includes.patch.gz
Description: Binary data
patch does
that.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center
diff --git a/src/backend/storage/lmgr/spin.c b/src/backend/storage/lmgr/spin.c
index 6d59a7f..353a3a9 100644
--- a/src/backend/storage/lmgr/spin.c
+++ b/src/backend/storage/lmgr/spin.c
@@ -22,13 +22,15 @@
*/
801 - 900 of 3221 matches
Mail list logo
