Re: VS 2015 support in src/tools/msvc

On 03/03/16 18:41, Magnus Hagander wrote:
> On Thu, Mar 3, 2016 at 6:18 PM, Andrew Dunstan <andrew(at)dunslane(dot)net
> <mailto:andrew(at)dunslane(dot)net>> wrote:
>
> On 03/03/2016 09:02 AM, Michael Paquier wrote:
>
> Microsoft provides a set of VMs that one can use for testing and
> Windows 10 is in the set:
> https://dev.windows.com/en-us/microsoft-edge/tools/vms/windows/
> I have grabbed one and installed the community version of Visual
> Studio 2015 so I think that I am going to be able to compile
> Postgres
> with VS2015 with a bit of black magic.
>
> So, would there be interest for a patch on the perl scripts in
> src/tools/msvc or are they considered a lost cause? Having a look at
> the failures could be done with the cmake work, but it seems a bit
> premature to me to look at that for the moment, and having
> support for
> VS2015 with 9.5 (support for new versions of VS won a backpatch the
> last couple of years) would be a good thing I think.
>
>
>
>
> I am not holding my breath on cmake. Until we have something pretty
> solid on that front I'm going to assume it's not happening. If we're
> going to support VS2015 (and I think we should) then it should be
> supported for all live branches if possible. I'm assuming the
> changes would be pretty localized, at least in src/tools/msvc, and
> adding a new compile shouldn't break anything with existing compilers.
>
>
> +1.
>
> Definitely do it for HEAD.
>
> Then if it gets backpatched is going to depend on the locality I think.
> If it's just the build system then it should be no problem, but I
> thought Michael also suggested some API changes. If that's so, then it
> is going to depend on how invasive those are. But that part should be
> done for HEAD regardless, so it's definitely worth the effort to figure
> out exactly what it involves.
>
>

Well the source code does not compile on MSVC2015, the perl changes
needed are really tiny, there is some code that needs changes to work
with 2015, particularly in the locale code-page detection area so it's
definitely not just build system. But I think it should be fairly
localized and fenced by ifdef anyway.

--
Petr Jelinek http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services