Date: Thu, 13 Jan 2022 17:58:27 -0600
From: Justin Pryzby <pryzby@telsasoft.com>
To: Daniil Zakhlystov <usernamedt@yandex-team.ru>
Cc: pgsql-hackers@lists.postgresql.org,
	Andrey Borodin <x4mmm@yandex-team.ru>
Subject: Re: libpq compression (part 2)
Message-ID: <20220113235827.GM14051@telsasoft.com>
References: <ABAA09C6-BB95-47A5-890D-90353533F9AC@yandex-team.ru>
 <20220112141526.GF14051@telsasoft.com>
 <D29C2E5B-30FA-4870-9CC8-D49DACDED7AF@yandex-team.ru>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <D29C2E5B-30FA-4870-9CC8-D49DACDED7AF@yandex-team.ru>
User-Agent: Mutt/1.9.4 (2018-02-28)
Archived-At: 
 <https://www.postgresql.org/message-id/20220113235827.GM14051%40telsasoft.com>
Precedence: bulk

On Fri, Jan 14, 2022 at 02:12:17AM +0500, Daniil Zakhlystov wrote:
> Hi, Justin!
> 
> First of all, thanks for the detailed review. I’ve applied your patches to the current version.

Note that my message had other comments that weren't addressed in this patch.

Your 0003 patch has a couple "noise" hunks that get rid of ^M characters added
in previous patches.  The ^M shouldn't be added in the first place.  Did you
apply my fixes using git-am or something else ?

On Fri, Jan 14, 2022 at 02:12:17AM +0500, Daniil Zakhlystov wrote:
> > On 12 Jan 2022, at 19:15, Justin Pryzby <pryzby@telsasoft.com> wrote:
> > 
> > zlib still causes check-world to get stuck.  I first mentioned this last March:
> > 20210319062800.GI11765@telsasoft.com
> > 
...
> > I removed the thread from the CFBOT until that's resolved.
> 
> I’ve fixed the failing tests, now they should pass.

macos: passed
linux: timed out after 1hr
freebsd: failed in pg_rewind: ... <= replay_lsn AND state = 'streaming' FROM ...
windows: "Failed test 'data_checksums=on is reported on an offline cluster stdout /(?^:^on$)/'" / WARNING:  01000: algorithm zlib is not supported

Note that it's possible and easy to kick off a CI run using any github account:
see ./src/tools/ci/README

For me, it's faster than running check-world -j4 locally, and runs tests on 4
OSes.

I re-ran your branch under my own account and linux didn't get stuck (and the
compiler warnings tests passed).  But on a third attempt, macos failed the
pg_rewind test, and bsd failed the subscription test:
| SELECT count(1) = 0 FROM pg_subscription_rel WHERE srsubstate NOT IN ('s', 'r');

-- 
Justin