Re: Backpatch src/test/modules to 9.3? - Mailing list pgsql-hackers

From Michael Paquier
Subject Re: Backpatch src/test/modules to 9.3?
Date
Msg-id CAB7nPqRmTb3jUGdT1fJeWw4Xuy68r3WQ5fxvQAusZE4qddYJkA@mail.gmail.com
Whole thread Raw
In response to Re: Backpatch src/test/modules to 9.3?  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
On Mon, Jun 22, 2015 at 1:48 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Andres Freund <andres@anarazel.de> writes:
>> When discussing multixacts around pgcon (including the developer
>> meeting) we decided that we want to add test code to make exercising the
>> truncation code easier.
>
>> In master the most logical place for that seems to be src/test/modules.
>> Unfortunately 9.3/9.4 don't have that. Given that the multixact code
>> isn't the same in the backbranches, and will hopefully diverge further,
>> I'd rather have tests there. So maybe we should just add
>> src/test/modules there? These then will not be exercised by the
>> buildfarm, but the affected code isn't particularly platform dependant,
>> so that doesn't worry me much?
>
> Presumably, once the files are there, it would be a simple adjustment
> to get the buildfarm to run those tests in branches >= 9.3 rather
> than >= 9.5.  So +1 for back-patching this infrastructure.

And let's not care about MSVC stuff in REL9_4_STABLE and REL9_3_STABLE
as well. The integration of src/test/modules has required some
refactoring of the MSVC scripts as those are basically extensions as
they needed to be identified as such in a path different than
contrib/, note as well that they are included in the installation to
facilitate their testing. And I don't think that we want to backpatch
this heavy refactoring in REL9_4_STABLE and REL9_3_STABLE for
stability's sake. Running them on other platforms is a good plan, so
+1 for having them, without diffs in src/tools/msvc of course. They
will be picked up by MinGW at least on Windows.
-- 
Michael



pgsql-hackers by date:

Previous
From: Michael Paquier
Date:
Subject: Re: SSL TAP tests and chmod
Next
From: Craig Ringer
Date:
Subject: Re: Time to get rid of PQnoPasswordSupplied?