In an earlier message, I said I didn't see the problem with xlc 13.1.3. I
withdraw that statement. I had tested old code (commit c629324, 20 Aug),
rendering the test invalid. xlc 13.1.3 does break commit 91d5f1a, and
removing "restrict" fixes things as it did the older version.
On Thu, Oct 12,
On 2017-10-12 19:35:36 -0700, Noah Misch wrote:
> On Thu, Oct 12, 2017 at 04:08:44PM -0700, Andres Freund wrote:
> > So we've two animals (hornet, sungazer) that are:
> > #define SIZEOF_SIZE_T 8
> > #define WORDS_BIGENDIAN 1
> > #define restrict __restrict
> >
> > one compiled with xlc that fails
Hi,
On 2017-10-13 00:02:47 -0700, Noah Misch wrote:
> I hacked psql to call PQtrace() and ran "psql -Xc 'select true'" in the
> defective configuration and in a working x64 GNU/Linux configuration. I've
> attached both PQtrace() products.
Thanks!
> To backend> Msg Q
> To backend> "select true"
Noah Misch writes:
> I hacked psql to call PQtrace() and ran "psql -Xc 'select true'" in the
> defective configuration and in a working x64 GNU/Linux configuration. I've
> attached both PQtrace() products.
Thanks. It looks to me like the xlc build simply forgets to send some
of the T-message fi
Rely on sizeof(typename) rather than sizeof(variable) in pqformat.h.
In each of the pq_writeintN functions, the three uses of sizeof() should
surely all be consistent. I started out to make them all sizeof(ni),
but on reflection let's make them sizeof(typename) instead. That's more
like our usua
Improve implementation of CRE-stack-flattening in map_variable_attnos().
I (tgl) objected to the obscure implementation introduced in commit
1c497fa72. This one seems a bit less action-at-a-distance-y, at the
price of repeating a few lines of code.
Improve the comments about what the function is
I wrote:
> Anyway, I will go make the sizeof() usages consistent, just to satisfy
> my own uncontrollable neatnik-ism. Assuming that hornet stays red,
> which is probable, we should disable "restrict" for that compiler.
As expected, that didn't fix it. Andres, are you going to put in
a disable?
On 2017-10-13 13:48:07 -0400, Tom Lane wrote:
> I wrote:
> > Anyway, I will go make the sizeof() usages consistent, just to satisfy
> > my own uncontrollable neatnik-ism. Assuming that hornet stays red,
> > which is probable, we should disable "restrict" for that compiler.
>
> As expected, that d
Andres Freund writes:
> A easiest way to do this would be to put something like
> CPPFLAGS="$CPPFLAGS -Dpg_restrict" into the existing
> if test "$GCC" != yes ; then
> block in a/src/template/linux. But that'd probably result in "macro
> redefined" warnings or somesuch.
You mean src/template/aix,
On 2017-10-13 14:19:22 -0400, Tom Lane wrote:
> > So it'd probably better to introduce a FORCE_DISABLE_RESTRICT=yes, set
> > at the same place, that's then tested before running the relevant
> > configure check?
>
> +1. I think you don't actually have to skip the configure check,
> and there migh
Fix possible crash with Parallel Bitmap Heap Scan.
If a Parallel Bitmap Heap scan's chain of leftmost descendents
includes a BitmapOr whose first child is a BitmapAnd, the prior coding
would mistakenly create a non-shared TIDBitmap and then try to perform
shared iteration.
Report by Tomas Vondra.
Fix possible crash with Parallel Bitmap Heap Scan.
If a Parallel Bitmap Heap scan's chain of leftmost descendents
includes a BitmapOr whose first child is a BitmapAnd, the prior coding
would mistakenly create a non-shared TIDBitmap and then try to perform
shared iteration.
Report by Tomas Vondra.
Force "restrict" not to be used when compiling with xlc.
Per buildfarm animal Hornet and followup manual testing by Noah Misch,
it appears xlc miscompiles code using "restrict" in at least some
cases. Allow disabling restrict usage with FORCE_DISABLE_RESTRICT=yes
in template files, and do so for a
Add pg_noinline macro to c.h.
Forcing a function not to be inlined can be useful if it's the
slow-path of a performance critical function, or should be visible in
profiles to allow for proper cost attribution.
Author: Andres Freund
Discussion:
https://postgr.es/m/20170914061207.zxotvyopetm7l...@
Improve sys/catcache performance.
The following are the individual improvements:
1) Avoidance of FunctionCallInfo based function calls, replaced by
more efficient functions with a native C argument interface.
2) Don't extract columns from a cache entry's tuple whenever matching
entries - ins
Andres Freund writes:
> Add pg_noinline macro to c.h.
I think you want this to be parenthesized so that pgindent doesn't
go nuts when you use it. At least, that was the reason why
pg_attribute_noreturn() has parens. Even if the current version
of pgindent doesn't have that problem, I would argu
On 2017-10-13 18:32:08 -0400, Tom Lane wrote:
> Andres Freund writes:
> > Add pg_noinline macro to c.h.
>
> I think you want this to be parenthesized so that pgindent doesn't
> go nuts when you use it. At least, that was the reason why
> pg_attribute_noreturn() has parens. Even if the current v
Add missing options to pg_regress help() output
A few command line options accepted by pg_regress were not being output
by help(), including --help itself. Add that one, as well as --version
and --bindir, and the corresponding short options for the first two.
We could consider this for backpatchi
18 matches
Mail list logo
