Re: GinPageIs* don't actually return a boolean

On Tue, Feb 9, 2016 at 11:53 PM, Michael Paquier
<michael(dot)paquier(at)gmail(dot)com> wrote:
> On Wed, Feb 10, 2016 at 1:17 AM, Yury Zhuravlev
> <u(dot)zhuravlev(at)postgrespro(dot)ru> wrote:
>> I've just run into a problem with these macro. Function ginStepRight breaks
>> completely when compiled using the MSVC2013 and MSVC2015 (since these
>> releases use C99's bools but without stdbool.h like C++).
>> I don't understand why the patch has not been commited yet. It seems to me
>> not so important !! or ! = 0, the solution is all that matters.
>
> +1 for applying it. There were some conflicts in gist and lwlock, that
> I fixed as per the attached. I have added as well an entry in next CF:
> https://commitfest.postgresql.org/9/507/
> If a committer wants to pick up that before, feel free. For now it
> won't fall in the void, that's better than nothing.

Not actually attached.

Are we thinking to back-patch this? I would be disinclined to
back-patch widespread changes like this. If there's a specific
instance related to Gin where this is causing a tangible problem, we
could back-patch just that part, with a clear description of that
problem. Otherwise, I think this should be master-only.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company