Re: pgsql: jit: Re-allow JIT compilation of execGrouping.c hashtable compar

Hi,

On September 29, 2019 4:34:14 PM PDT, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>Andres Freund <andres(at)anarazel(dot)de> writes:
>> jit: Re-allow JIT compilation of execGrouping.c hashtable
>comparisons.
>
>I've got to say that I'm not very comfortable with changing stuff
>when we've got less than 24 hours to the 12.0 GA wrap.

I asked for input on hackers, nobody replied... I noticed the l the speed difference when doing casual 11 vs 12 testing (with an old v11 build, leading me to discover both issues), so it seems quite likely that others would be affected too.

> At this
>point you're not going to get complete buildfarm feedback (notably,
>not from CLOBBER_CACHE_ALWAYS animals). Unless it's a horrible
>crasher bug, it'd be better to wait till post-tag and ship it in
>12.1.

There's no animal doing CCA or valgrind with JIT enabled anyway... And the changes have no meaningful effect outside of jit (which I plan to change soon in master, but that obviously doesn't matter).

Andres
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.