Re: Recovery mode with partitioned tables and ANY(NULL) in WHERE clause

piotrwlodarczyk89@gmail.com

From: Piotr Włodarczyk <piotrwlodarczyk89@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: pgsql-bugs@lists.postgresql.org
Date: 2019-08-09T17:44:00Z
Lists: pgsql-bugs
Hi,

I've checked your commit and let me ensue it's fulfill all cases.
For example what with query like that: select * from coercepart where a =
any(null); ? there is no casting to array of specified type. I'm not master
in C but looking in source I believe that "if (array->const snull)" is
enough. But for better check you can add that query (and similar) to
tests/regress



On Fri, Aug 9, 2019 at 7:23 PM Piotr Włodarczyk <piotrwlodarczyk89@gmail.com>
wrote:

> Cool, thanks!
>
> pt., 9 sie 2019, 19:21 użytkownik Tom Lane <tgl@sss.pgh.pa.us> napisał:
>
>> =?UTF-8?Q?Piotr_W=C5=82odarczyk?= <piotrwlodarczyk89@gmail.com> writes:
>> > Can you tell me in which versions it'll be fixed? For me it's critical
>> > situation because we can have some potentially dangerous places in
>> > application.
>>
>> Unfortunately this just missed this week's releases, so it'll be the
>> next quarterly releases in November.
>>
>> If you're sufficiently worried about it you could apply the patch locally.
>> It's pretty trivial:
>>
>>
>> https://git.postgresql.org/pg/commitdiff/2f729d83226705d1149419a2aef7c1678fe641ec
>>
>>                         regards, tom lane
>>
>

-- 

Pozdrawiam
Piotr Włodarczyk

Commits

  1. Fix SIGSEGV in pruning for ScalarArrayOp with constant-null array.