On Mon, Aug 29, 2022 at 3:19 PM John Naylor
wrote:
>
> It turns out MSVC animal drongo doesn't like this cast -- on x86 they
> are the same underlying type. Will look into that as more results come
> in.
Here's the simplest fix I can think of:
/*
* Exactly like vector8_is_highbi
> mb/pg_wchar.h include out to before we include all the perl stuff. That does
> fix the issue.
We could also move is_valid_ascii somewhere else. It's only
tangentially related to "wide chars" anyway.
--
John Naylor
EDB: http://www.enterprisedb.com
and without
the patch, to see how the inflection point shifts. That worked pretty
well when testing dual-pivot.
--
John Naylor
EDB: http://www.enterprisedb.com
From fd4bfdbee88478fac32838712c112d91f73c5db5 Mon Sep 17 00:00:00 2001
From: John Naylor
Date: Wed, 4 May 2022 23:35:06 +0700
Subject: [PATCH
On Mon, Aug 29, 2022 at 12:44 PM Nathan Bossart
wrote:
> [v6]
Pushed with a couple comment adjustments, let's see what the build
farm thinks...
--
John Naylor
EDB: http://www.enterprisedb.com
meliorated. As I said in the dual pivot
thread, this would be great for dual pivot since we could reuse
partial insertion sort for choosing the pivots, reducing binary space.
--
John Naylor
EDB: http://www.enterprisedb.com
wonder if we should explain briefly what saturating arithmetic is. I
had never encountered it outside of a SIMD programming context.
--
John Naylor
EDB: http://www.enterprisedb.com
On Mon, Aug 29, 2022 at 11:25 AM John Naylor
wrote:
> +static inline bool
> +vector32_is_highbit_set(const Vector32 v)
> +{
> +#ifdef USE_SSE2
> + return (_mm_movemask_epi8(v) & 0x) != 0;
> +#endif
> +}
>
> I'm not sure why we need this function -- AFAICS
On Mon, Aug 29, 2022 at 4:28 PM John Naylor
wrote:
>
> Here's the simplest fix I can think of:
>
> /*
> * Exactly like vector8_is_highbit_set except for the input type, so
> it still looks
> * at each _byte_ separately.
> *
> * XXX x86 uses the same underlying
Putting the
> include in plperl.h would keep that aspect transparent, because plperl_utils.h
> includes plperl.h.
Since plperl_helpers.h already includes plperl.h, I'm not sure why
both are included everywhere the former is. If .c/.xs files didn't
include plperl.h directly, we could keep pg_wchar.h in
plperl_helpers.h. Not sure if that's workable or any better...
--
John Naylor
EDB: http://www.enterprisedb.com
On Sat, Aug 27, 2022 at 10:02 AM Tom Lane wrote:
>
> John Naylor writes:
> > On Sat, Aug 27, 2022 at 4:15 AM Andres Freund wrote:
> >> I think it'd also be needed in hstore_plperl.c, jsonb_plperl.c. Putting the
> >> include in plperl.h would keep th
cast(c));
+ result = vector8_has_zero(sub);
I think we should invent a helper that does saturating subtraction and
call that, inlining the sub var so we don't need to mess with it
further.
Otherwise seems fine.
--
John Naylor
EDB: http://www.enterprisedb.com
On Sat, Aug 27, 2022 at 11:20 AM John Naylor
wrote:
>
> Here's a patch with that idea, not tested on Windows yet.
Update: I tried taking the CI for a spin, but ran into IT issues with
Github when I tried to push my branch to remote.
--
John Naylor
EDB: http://www.enterprisedb.com
On Sat, Aug 27, 2022 at 2:23 PM Andres Freund wrote:
>
> Hi,
>
> On 2022-08-27 12:53:24 +0700, John Naylor wrote:
> > Update: I tried taking the CI for a spin, but ran into IT issues with
> > Github when I tried to push my branch to remote.
>
> A github, not a CI is
On Tue, Aug 30, 2022 at 12:17 AM Nathan Bossart
wrote:
> Thanks! I've attached a follow-up patch with a couple of small
> suggestions.
Pushed, thanks!
--
John Naylor
EDB: http://www.enterprisedb.com
onal flags?
I'm wondering if __aarch64__ would be clearer on that, and if we get
windows-on-arm support as has been proposed, could also add _M_ARM64.
I also see #if defined(__aarch64__) || defined(__aarch64) in our
codebase already, but I'm not sure what recognizes the latter.
--
John Naylor
EDB: http://www.enterprisedb.com
it has a different API,
requires new functions we don't need, and possibly has portability
issues. A simple for-loop is better for assertions.
- A runtime elog is not appropriate for a compile time check -- use
#error instead.
--
John Naylor
EDB: http://www.enterprisedb.com
On Tue, Aug 23, 2022 at 1:03 PM John Naylor
wrote:
>
> LGTM overall. My plan is to split out the json piece, adding tests for
> that, and commit the infrastructure for it fairly soon.
Here's the final piece. I debated how many tests to add and decided it
was probably enough to add
GHT-TO-LEFT MARK;Cf;0;R;N;
So maybe we need to take Cf characters in this file into account, in
addition to Me and Mn (combining characters).
--
John Naylor
EDB: http://www.enterprisedb.com
ome buildfarm animals recently, it seems the
oldest perl there is 5.14? ... which came out in 2011, so it seems
later on we could just set that as the minimum.
--
John Naylor
EDB: http://www.enterprisedb.com
On Fri, Sep 2, 2022 at 12:17 PM Kyotaro Horiguchi
wrote:
>
> At Thu, 01 Sep 2022 18:22:06 +0900 (JST), Kyotaro Horiguchi
> wrote in
> > At Thu, 1 Sep 2022 15:00:38 +0700, John Naylor
> > wrote in
> > > UnicodeData.txt has this:
> > >
> > > 200
On Wed, Aug 31, 2022 at 11:17 AM Nathan Bossart
wrote:
>
> On Wed, Aug 31, 2022 at 10:50:39AM +0700, John Naylor wrote:
> > Here's the final piece. I debated how many tests to add and decided it
> > was probably enough to add one each for checking quotes and
> > back
header
> > Subject: [PATCH v4 03/11] Build guc-file.c standalone
>
> 01-03 are a bit more complicated, but still look not far off. There's a FIXME
> about failing headercheck.
--
John Naylor
EDB: http://www.enterprisedb.com
r (int i = 0; i < sizeof(Vector8); i++)
> | ^
"int" should probably be "Size" -- does that remove the warning?
--
John Naylor
EDB: http://www.enterprisedb.com
On Sat, Sep 3, 2022 at 12:57 PM Tom Lane wrote:
>
> John Naylor writes:
> > On Sat, Sep 3, 2022 at 12:30 PM Pavel Stehule
> > wrote:
> >> /usr/local/pgsql/master/include/server/port/simd.h: In function
> >> ‘vector8_has’:
> >> /usr/local/pgsql/mast
o how I
did json path in the last patch. Also removed the bogus include from
v4 to . Pushed 01 and 02 separately, then squashed and pushed the
rest.
--
John Naylor
EDB: http://www.enterprisedb.com
On Fri, Sep 2, 2022 at 11:35 PM Andres Freund wrote:
>
> Hi,
>
> On 2022-09-02 14:17:26 +0700, John Naylor wrote:
> > On Thu, Sep 1, 2022 at 1:12 AM Andres Freund wrote:
> > > [v12]
> >
> > +# Build a small utility static lib for the parser. This ma
arning".
>
> A positive argument for doing nothing is that there's room to worry
> whether -D__STDC_WANT_LIB_EXT1__ might have any side-effects we
> *don't* want.
This is still listed in the CF as needing review, so I went and marked
it rejected.
--
John Naylor
EDB: http://www.enterprisedb.com
On Mon, Sep 5, 2022 at 1:18 AM Andres Freund wrote:
>
> Hi,
>
> On 2022-09-04 12:16:10 +0700, John Naylor wrote:
> > Pushed 01 and 02 separately, then squashed and pushed the rest.
>
> Thanks a lot! It does look a good bit cleaner to me now.
>
> I think, as a follow
Here are autoconf-only patches to that effect.
--
John Naylor
EDB: http://www.enterprisedb.com
From 5e07442ebf728c486536038aecf29aefcc892455 Mon Sep 17 00:00:00 2001
From: John Naylor
Date: Tue, 6 Sep 2022 11:41:58 +0700
Subject: [PATCH v2 2/2] Bump minimum version of Flex to 2.5.35
Since the retirement o
On Mon, Sep 5, 2022 at 4:11 AM Andres Freund wrote:
>
> Hi,
>
> On 2022-09-04 13:12:52 +0700, John Naylor wrote:
> > On Fri, Sep 2, 2022 at 11:35 PM Andres Freund wrote:
> > >
> > > Hi,
> > >
> > > On 2022-09-02 14:17:26 +0700, John Naylo
On Tue, Aug 23, 2022 at 4:15 AM Nathan Bossart wrote:
>
> On Mon, Aug 22, 2022 at 11:50:35AM +0700, John Naylor wrote:
> > Is this also ever defined on 32-bit? If so, is it safe, meaning the
> > compiler will not emit these instructions without additional flags?
> > I'm w
On Wed, Aug 24, 2022 at 12:15 AM Nathan Bossart
wrote:
>
> On Tue, Aug 23, 2022 at 01:03:03PM +0700, John Naylor wrote:
> > On Tue, Aug 23, 2022 at 10:32 AM Nathan Bossart
> >> Here's a new version of the patch with the 32-bit changes and calls to
> >> lfind() rem
On Tue, Aug 23, 2022 at 7:14 AM Peter Smith wrote:
>
> I noticed an accidental ;;
>
> PSA patch to remove the same.
Pushed.
--
John Naylor
EDB: http://www.enterprisedb.com
, and (only if there
are additional sort keys) once on the null partition. And the
pre-partitioning would take care of nulls first/last upfront. I
haven't looked into the feasibility of this yet, but the good thing
about the concept is that it removes null handling in the comparators
without addition
On Tue, Aug 23, 2022 at 10:32 AM Nathan Bossart
wrote:
>
> On Mon, Aug 22, 2022 at 02:22:29PM -0700, Nathan Bossart wrote:
> > On Mon, Aug 22, 2022 at 09:35:34AM +0700, John Naylor wrote:
> >> Not at all! However, the 32-bit-element changes are irrelevant for
> >&
On Tue, Aug 23, 2022 at 11:24 AM David Rowley wrote:
>
> On Tue, 23 Aug 2022 at 15:22, John Naylor
> wrote:
> > Did you happen to see
> >
> > https://www.postgresql.org/message-id/CAFBsxsFhq8VUSkUL5YO17cFXbCPwtbbxBu%2Bd9MFrrsssfDXm3Q%40mail.gmail.com
>
> I m
On Thu, Aug 25, 2022 at 11:57 AM Nathan Bossart
wrote:
>
> On Thu, Aug 25, 2022 at 10:38:34AM +0700, John Naylor wrote:
> > On Thu, Aug 25, 2022 at 1:01 AM Nathan Bossart
> > wrote:
> >> On Wed, Aug 24, 2022 at 11:07:03AM +0700, John Naylor wrote:
> >> > -
"==" instead of "<=" for this one?
Good catch, will fix.
--
John Naylor
EDB: http://www.enterprisedb.com
On Fri, Aug 26, 2022 at 10:26 AM Nathan Bossart
wrote:
>
> On Thu, Aug 25, 2022 at 04:41:53PM +0700, John Naylor wrote:
> > v3 applies on top of the v9 json_lex_string patch in [1] and adds a
> > bit more to that, resulting in a simpler patch that is more amenable
> > to
On Thu, Aug 25, 2022 at 1:35 PM John Naylor
wrote:
>
> I think I'll go ahead and commit 0001 in a couple days pending further
> comments.
Pushed with Nathan's correction and some cosmetic rearrangements.
--
John Naylor
EDB: http://www.enterprisedb.com
On Tue, Aug 16, 2022 at 4:23 AM Nathan Bossart wrote:
>
> On Mon, Aug 15, 2022 at 08:33:21PM +0700, John Naylor wrote:
> > +#ifdef USE_SSE2
> > + chunk = _mm_loadu_si128((const __m128i *) [i]);
> > +#else
> > + memcpy(, [i
On Wed, Sep 28, 2022 at 1:18 PM John Naylor
wrote:
> [stuff about size classes]
I kind of buried the lede here on one thing: If we only have 4 kinds
regardless of the number of size classes, we can use 2 bits of the pointer
for dispatch, which would only require 4-byte alignment. That sho
On Tue, Sep 27, 2022 at 2:06 AM Andres Freund wrote:
>
> On 2022-09-26 15:18:29 +0700, John Naylor wrote:
> > Either way it doesn't exist on this machine. I was able to get a working
> > build with
> >
> > /usr/local/Homebrew/Library/Homebrew/os/mac/pkgconfig
>
&
at the behavior is.
> I also took the chance to simplify the code a little bit.
It's perfectly clear and simple now, even if it doesn't win at "code golf".
--
John Naylor
EDB: http://www.enterprisedb.com
On Tue, Oct 25, 2022 at 9:48 AM Richard Guo wrote:
>
>
> On Tue, Oct 25, 2022 at 10:05 AM John Naylor
wrote:
>>
>> It's perfectly clear and simple now, even if it doesn't win at "code
golf".
>
>
> Agree with your point. Do you think we can furt
This would have to change when we implement shrinking of nodes, but might
still be useful.
+ if (!rt_node_search_leaf(node, key, RT_ACTION_FIND, value_p))
+ return false;
+
+ return true;
Maybe just "return rt_node_search_leaf(...)" ?
--
John Naylor
EDB: http://www.enterprisedb.com
info_elem rt_node_kind_info[RT_NODE_KIND_COUNT] = {
In the assembly, I see it checks if there is room in the node by doing a
runtime lookup in this array, which is not constant. This might not be
important just yet, because I want to base the check on the proposed node
capacity instead, but I menti
a naive user -- it would be good to try
running everything documented here.
--
John Naylor
EDB: http://www.enterprisedb.com
On Thu, Oct 20, 2022 at 1:55 AM Andres Freund wrote:
>
> Hi,
>
> On 2022-10-11 10:21:17 +0700, John Naylor wrote:
> > On Tue, Oct 11, 2022 at 5:31 AM David Rowley
wrote:
> > >
> > > The proposed patches in [1] do aim to make additional usages of the
> &
homebrew section is still labeled with "find MacPorts
libraries".
--
John Naylor
EDB: http://www.enterprisedb.com
On Tue, Sep 20, 2022 at 3:19 PM Masahiko Sawada
wrote:
>
> On Fri, Sep 16, 2022 at 4:54 PM John Naylor
> wrote:
> > Here again, I'd rather put this off and focus on getting the "large
> > details" in good enough shape so we can got towards integrating
he actual logic/data change is low-risk. The patch renames
two files, which seems a bit much this late in the cycle. Maybe that's
okay, but I'd like someone else to opine before doing so.
--
John Naylor
EDB: http://www.enterprisedb.com
On Fri, Sep 9, 2022 at 12:18 PM John Naylor
wrote:
>
> It seems gramparse.h isn't installed now? In any case, here's a patch
> to move gramparse to the backend dir and stop symlinking/ installing
> gram.h.
Looking more closely at src/include/Makefile, this is incorrect -- all
file
older Windows and Visual
Studio versions, maybe it's low-enough risk defer the check to the
Meson conversion? I understand our MSVC process will then go away much
more quickly than autoconf...
--
John Naylor
EDB: http://www.enterprisedb.com
From d8e241968395c552d667c1eb393cbaea12f0df6c Mon Sep 17 00
On Wed, Sep 14, 2022 at 10:46 AM Tom Lane wrote:
>
> John Naylor writes:
> > On Wed, Sep 14, 2022 at 6:47 AM Tom Lane wrote:
> >> I've just switched longfin to use built-from-source perl 5.14.0.
>
> > In that case, here is a quick update with commit message. N
On Wed, Sep 14, 2022 at 6:10 AM Andres Freund wrote:
>
> On 2022-09-12 14:49:50 +0700, John Naylor wrote:
> > CI builds fine. For v2 I only adjusted the commit message. I'll push
> > in a couple days unless there are objections.
>
> Makes sense to me. Thanks for work
fair, our grammar changes much more often. One other possible
deal-breaker of that is that it makes it more painful for forks to
maintain additional syntax.
--
John Naylor
EDB: http://www.enterprisedb.com
m size, so there
is no up front calculation assuming max-tuples-per-page, so it
automatically uses less memory for less demanding tables.
(But +1 for changing that calculation for as long as we do have the
single array.)
--
John Naylor
EDB: http://www.enterprisedb.com
rker without shared
memory access are not supported",
+ errmsg("background worker \"%s\": background workers without shared
memory access are not supported",
is a grammar error so worth backpatching, but the rest are cosmetic.
Will commit this way unless there are objections.
--
John Naylor
EDB: http://www.enterprisedb.com
so we can say
> "5.14" in the docs without fine print.
Until such time as that happens, here is a draft to require 5.14.2.
--
John Naylor
EDB: http://www.enterprisedb.com
config/perl.m4 | 4 ++--
configure| 6 +++---
doc/src/sgm
> readable. Although for this case, a handy written parser was trivial too.
If the hand-written version is trivial, then we should prefer it.
--
John Naylor
EDB: http://www.enterprisedb.com
On Thu, Sep 8, 2022 at 12:39 PM John Naylor
wrote:
>
> On Fri, Sep 2, 2022 at 3:19 PM Kyotaro Horiguchi
> wrote:
> >
> > At Fri, 2 Sep 2022 13:43:50 +0700, John Naylor
> > wrote in
> > > If there is any doubt about including all of Cf, we could also just
&
On Tue, Sep 13, 2022 at 5:53 PM John Naylor
wrote:
>
> Until such time as that happens, here is a draft to require 5.14.2.
As soon as I hit send, it occurred to me that we don't check the perl
version on Windows, since (I seem to recall) 5.8.3 was too old to be
an option on that platform.
On Fri, Sep 2, 2022 at 3:19 PM Kyotaro Horiguchi
wrote:
>
> At Fri, 2 Sep 2022 13:43:50 +0700, John Naylor
> wrote in
> > If there is any doubt about including all of Cf, we could also just
> > add a branch in wchar.c to hard-code the 200B-200F range.
>
> If every wa
On Wed, Sep 7, 2022 at 3:36 PM Alvaro Herrera wrote:
>
> On 2022-Sep-06, John Naylor wrote:
>
> > Note that the indentation hasn't changed. My thought there: perltidy
> > will be run again next year, at which time it will be part of a listed
> > whitespace-only co
>
> I found in my notes:
>
> * maybe gram.h and gramparse.h should not be installed
>
> So, yeah. ;-)
It seems gramparse.h isn't installed now? In any case, here's a patch
to move gramparse to the backend dir and stop symlinking/ installing
gram.h. Confusingly, MSVC didn't seem to copy gram.h t
On Fri, Sep 9, 2022 at 12:07 AM Andres Freund wrote:
>
> Hi,
>
> On 2022-09-06 13:32:32 +0700, John Naylor wrote:
> > Here are autoconf-only patches to that effect.
>
> Looks like you did actually include src/tools/msvc as well :)
Ah, in my head I meant "no patches f
AM Thomas Munro wrote:
> Remaining things from this thread:
> * removing --disable-thread-safety
> * removing those vestigial HAVE_XXX macros (one by one analysis and patches)
> * making Unix sockets secure for Windows in tests
--
John Naylor
EDB: http://www.enterprisedb.com
ndamental rethinking of memory management, error handling,
> and some other things". That's pretty much a non-starter.
Added "Rewrite the code in a different language" to "Features We Do Not
Want" section of Wiki, referencing the two threads that came up:
htt
On Fri, Sep 16, 2022 at 10:51 AM John Naylor
wrote:
>
> On Fri, Sep 16, 2022 at 10:11 AM Kyotaro Horiguchi
> wrote:
> >
> > The patch seems to me covering all occurances of PG_PROC as PGPROC.
>
> +1 since this hinders grep-ability.
Pushed this.
> >
On Thu, Sep 22, 2022 at 7:52 PM John Naylor
wrote:
>
>
> On Thu, Sep 22, 2022 at 1:26 PM Masahiko Sawada
wrote:
> > Good point. While keeping the chunks in the small nodes in sorted
> > order is useful for visiting all keys in sorted order, additional
> > branches and
On Thu, Sep 22, 2022 at 11:46 AM John Naylor
wrote:
> One thing I want to try soon is storing fewer than 16/32 etc entries, so
that the whole node fits comfortably inside a power-of-two allocation. That
would allow us to use aset without wasting space for the smaller nodes,
which would be fas
On Thu, Sep 22, 2022 at 1:26 PM Masahiko Sawada
wrote:
>
> On Thu, Sep 22, 2022 at 1:46 PM John Naylor
> wrote:
> > While on the subject, I wonder how important it is to keep the chunks
in the small nodes in sorted order. That adds branches and memmove calls,
and is t
to get a working build with
/usr/local/Homebrew/Library/Homebrew/os/mac/pkgconfig
(My homebrew install doesn't seem to have anything relevant for
extra_include_dirs or extra_lib_dirs.)
2) Also, "ninja -v install" has the same line count as "ninja install" --
are there versions t
On Fri, Sep 16, 2022 at 1:01 PM Masahiko Sawada wrote:
>
> On Mon, Aug 15, 2022 at 10:39 PM John Naylor
> wrote:
> >
> > bool, buth = and <=. Should be pretty close. Also, i believe if you
> > left this for last as a possible refactoring, it might save some work.
v
overing all occurances of PG_PROC as PGPROC.
+1 since this hinders grep-ability.
> I found several uses of PG_PROC as (pg_catalog.)pg_proc, which is
> quite confusing, too..
It's pretty obvious to me what that refers to in primnodes.h, although
the capitalization of (some, but not all) catalo
On Wed, Sep 14, 2022 at 5:25 PM John Naylor
wrote:
> Will commit this way unless there are objections.
I forgot to mention yesterday, but this is done.
--
John Naylor
EDB: http://www.enterprisedb.com
On Wed, Sep 14, 2022 at 11:33 PM Peter Geoghegan wrote:
>
> On Wed, Sep 14, 2022 at 3:18 AM John Naylor
> > Furthermore, it doesn't have to anticipate the maximum size, so there
> > is no up front calculation assuming max-tuples-per-page, so it
> > automatically u
ings worse in this respect. Do you think we should address this
> too? I suspect that describing the difference between the header and
> the type tag here will create even more confusion.
I said nothing about describing the difference between the header and
type tag. The patch added xxx's for the type tag in a comment about
the header. This is more misleading than what is there now.
--
John Naylor
EDB: http://www.enterprisedb.com
.)
>
> The approach with the make targets is better than a configure option,
> because it allows you to build a narrow set of things during development
> and then build everything for final confirmation, without having to
> re-run configure. Also, it's less confusing for packagers.
Another point is that the --with-FOO options are intended for building
and linking with external library FOO.
--
John Naylor
EDB: http://www.enterprisedb.com
uce derivation
Filters
↳ 2: Filters Filter
↳ 1: ε • ↳ 5: C_EXCLUDE exclude_object pattern
--
John Naylor
EDB: http://www.enterprisedb.com
.
>
>
> I think it can be fixed just in v15 and master. This issue has no impact on
> SQL.
Well, if the regressions from v11 are not important enough to
backpatch, there is not as much of a case to backpatch the full fix to
v15 either.
--
John Naylor
EDB: http://www.enterprisedb.com
ader is at
a kind of "local maximum" -- minor adjustments are more likely to make
it worse. To significantly improve clarity might require a larger
rewriting, but I'm not personally interested in taking part in that.
--
John Naylor
EDB: http://www.enterprisedb.com
On Thu, Sep 22, 2022 at 1:01 AM Nathan Bossart
wrote:
>
> On Wed, Sep 21, 2022 at 01:17:21PM +0700, John Naylor wrote:
>
> > In short, this code needs to be lower level so that we still have full
> > control while being portable. I will work on this, and also the rela
On Tue, Sep 27, 2022 at 2:41 PM John Naylor
wrote:
>
> On Tue, Sep 27, 2022 at 2:06 AM Andres Freund wrote:
> >
> > On 2022-09-26 15:18:29 +0700, John Naylor wrote:
> > Yea, it's /usr/local on x86-64, based on what was required to make
macos CI
> > work. I updat
On Wed, Oct 5, 2022 at 1:46 PM Masahiko Sawada
wrote:
>
> On Wed, Sep 28, 2022 at 12:49 PM Masahiko Sawada
wrote:
> >
> > On Fri, Sep 23, 2022 at 12:11 AM John Naylor
> > wrote:
> > Yeah, node31 and node256 are bloated. We probably could use slab for
> &g
On Thu, Oct 6, 2022 at 2:53 PM Masahiko Sawada
wrote:
>
> On Wed, Oct 5, 2022 at 6:40 PM John Naylor
wrote:
> >
> > This wasn't the focus of your current email, but while experimenting
with v6 I had another thought about local allocation: If we use the default
slab block
I have proposed in that thread a scheme to squeeze
things into power-of-two sizes without wasting quite as much space. That's
not a done deal, of course, but it could work today without adding memory
management code.
[1]
https://www.postgresql.org/message-id/20220704220038.at2ane5xkymzzssb%40awork3.anarazel.de
On Sat, Sep 17, 2022 at 12:29 PM Nathan Bossart
wrote:
>
> On Fri, Sep 16, 2022 at 02:54:14PM +0700, John Naylor wrote:
> > v6 demonstrates why this should have been put off towards the end.
(more below)
>
> Since the SIMD code is fresh in my mind, I wanted to offer my
nds on SSE4.2. We can't assume those
currently.
--
John Naylor
EDB: http://www.enterprisedb.com
ple: we don't shrink at all now. :-)
--
John Naylor
EDB: http://www.enterprisedb.com
On Mon, Oct 10, 2022 at 12:16 PM John Naylor
wrote:
> Thanks for that! Now I can show clear results on some aspects in a simple
way. The attached patches (apply on top of v6)
Forgot the patchset...
--
John Naylor
EDB: http://www.enterprisedb.com
radix-v6-addendum-jcn1.tar.gz
Descript
> You mean semi-colon? +1, and a quick regex later I don't see any other
> occurrence.
Interestingly, I did find that in C, some statement is needed after a
label, even an empty one, otherwise it won't compile. That's not the case
here, though, so pushed.
--
John Naylor
EDB: http://www.enterprisedb.com
e serious about trying to
improve insertion sort performance, the simple idea we discussed at the
start of the thread is a much more modest step that has a good chance of
justifying the time put into it. That is not to say it's easy, however,
because testing is a non-trivial amount of work.
--
air, but it's usually
small compared to array size. Somewhat unrelated, we could still implement
Andres' idea [1] to dispense with the isset array in inner nodes of the
indirect array type (now node128), since we can just test if the pointer is
null.
[1]
https://www.postgresql.org/message-id/20220704
array_serach_ms
-+--+-++---+--+-
100 | 10199040 | 18000 |171 |
107 | 827 | 1400
--
John Naylor
EDB: http://www.enterprisedb.com
From 43a50a385930ee340d0a3b003910c704a0ff342
On Thu, Aug 4, 2022 at 8:41 PM Tom Lane wrote:
>
> John Naylor writes:
> > RepOriginId is a typedef for uint16, so this can't print the wrong
answer,
> > but it is inconsistent with other uses. So it seems we don't need to
> > backpatch this one?
>
> Um ..
On Fri, Aug 5, 2022 at 5:15 AM Nathan Bossart
wrote:
>
> On Thu, Aug 04, 2022 at 02:58:14PM +0700, John Naylor wrote:
> > Were you considering adding the new function to simd.h now that that's
> > committed? It's a bit up in the air what should go in there, but this
new
> &g
of code and it's difficult to see what's different for
each platform, IMO.
--
John Naylor
EDB: http://www.enterprisedb.com
From 69d56a21192ed2f03bc08f078cfff7ba5cb0d80b Mon Sep 17 00:00:00 2001
From: John Naylor
Date: Fri, 5 Aug 2022 13:30:47 +0700
Subject: [PATCH v2] Use SSE2 in is_valid_ascii wher
(). Here
> is an example.
LGTM, let's see what the buildfarm thinks of 0001.
--
John Naylor
EDB: http://www.enterprisedb.com
801 - 900 of 1330 matches
Mail list logo