Horton

Re: MSVC buildfarm critters are not running modules' TAP tests

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>, pgsql-hackers@lists.postgresql.org
Date: 2019-09-09T00:43:06Z
Lists: pgsql-hackers 
On Sun, Sep 08, 2019 at 07:44:16PM -0400, Tom Lane wrote:
> Yeah, perhaps, but I wasn't sure about how to do that easily.
> Feel free to add it ...

Feeding the makefile contents and then doing a lookup using =~ should
be enough.  I think that we should just refactor set of routines for
fetchTests() so as it uses the flag to look after as input.
fetchRegressOpts() has tweaks for ENCODING and NO_LOCALE which don't
apply to the isolation tests so perhaps a different routine would be
better.

> Hm, yeah, if there are any such tests in those directories.

snapshot_too_old is a good example to work with.

> The whole thing should be back-patched into branches that have
> any affected tests.  (But, please, not till after beta4 is
> tagged.)

Sure.  Don't worry about that.  I am focused on another thing lately
and it does not touch back-branches.
--
Michael

Commits

  1. Expand properly list of TAP tests used for prove in vcregress.pl

  2. Avoid using INFO elevel for what are fundamentally debug messages.