2015-01-06 12:22 GMT+01:00 Jacek Caban <jacek(a)codeweavers.com>:
On 01/05/15 21:58, Erik van Pienbroek wrote:
> Jacek Caban schreef op ma 05-01-2015 om 14:05 [+0100]:
>> On 01/04/15 12:49, Jacek Caban wrote:
>>> Maybe I missed some better options for us. None of above is perfect and
>>> I'm not sure what we should do about it. Solution 2. seems the least
>>> problematic.
>> Looking deeper at this, current implementation has one more problem. We
>> can't really have localtime_r, because it needs to depend on
>> _USE_32BIT_TIME_T macro. So if we really wanted to have a real function
>> in mingwex, we'd need it as localtime32_r and localtime64_r and an
>> inline wrapper. Given that, I think we should live with inline
>> implementation. Esp. since we may use localtime_s (which already has
>> wrapper inline as well as compatibility stub in libmsvcrt.a), which
>> makes the implementation trivial. Please review the attached patch. I
>> believe we should do the same for ctime_r and asctime_r.
> Hi Jacek,
>
> Thanks for the patch. I just tested it and I can confirm that it solves
> the localtime_r issue in glib2 and the gmtime_r issues in libgsf and
> libsoup.
Thanks for testing. Kai, what do you think, should I commit the patch?
Sure. Patch is ok.
> The cmtime_r issue in cairo is not resolved yet with this
> patch, but I guess this is expected for now.
Yeah, I may prepare a patch for that as well if we decided to go this way.
This route makes sense IMO. So what are others thinking about it?
> However, there are now other issues which prevent glib2 from
building
> successfully (libgsf and libsoup built just fine).
>
> For glib2:
> ../../gio/gsocket.c:1934:1: error: conflicting types for
> 'if_nametoindex'
> if_nametoindex (const gchar *iface)
> ^
> In file included
> from /usr/i686-w64-mingw32/sys-root/mingw/include/iphlpapi.h:16:0,
> from ./gnetworking.h:35,
> from ../../gio/gnetworkingprivate.h:22,
> from ../../gio/gsocket.c:60:
> /usr/i686-w64-mingw32/sys-root/mingw/include/netioapi.h:321:20: note:
> previous declaration of 'if_nametoindex' was here
> NET_IFINDEX WINAPI if_nametoindex(
> ^
>
> My first guess is that this needs to be fixed in glib2 itself and that
> it is unrelated to your patch
Yeah, that would be my guess too.
Yes, I would assume so too.
Kai