On 04/26/2018 01:29 AM, Michael Paquier wrote:
> On Thu, Apr 26, 2018 at 09:06:09AM +0500, Haroon wrote:
>> Apparently the following patch[1] got committed to head only (9.6 at the
>> time) and never made it into back branches i.e. 9.5, 9.4 and 9.3. Is it an
>> oversight ?
>>
>> [1]
>> http://git.postgresql.org/pg/commitdiff/868628e4fd44d75987d6c099ac63613cc5417629
> The support of VS 2015 being back-patched down to 9.5 is intentional,
> as the ideawhen integrating a new integration with visual studio is to
> patch HEAD plus the latest major version already released, at least
> that's the move respected for the last couple of years.
>
> Now I think that you are right: we should have this patch present as
> well in REL9_5_STABLE in order to be able to get the combination
> REL9_5_STABLE/VS2015 working correctly. At least we have made efforts
> to claim support of VS2015 down to 9.5, so I would suggest to
> cherry-pick it and make the solution complete.
>
> The discussion about _timezone and _tzdata is here by the way:
> https://www.postgresql.org/message-id/CAB7nPqTEkqF7E7nWcdQSvyOUAV3OGjX%3DrKSxFTkk03TOJGF5ng%40mail.gmail.com
>
> There is dory in the buildfarm which compiles using VS 2015, but it runs
> only 9.6 and newer versions.
>
> Robert? Or Magnus perhaps?
Well, the commit suggests that we should be guided by what the buildfarm
did, and then we never followed that up. I'd say in the light of
experience we should backpatch it to the remaining live branches.
cheers
andrew
--
Andrew Dunstan https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services