Le 2015-07-01 15:35, wm4 a écrit :
We were discussing some necessary library-safety cleanups on
#libav-devel. One of them is the removal of the "lock manager", which
we
all agreed that it must die. This would be pretty simple with
pthreads
and modern win32 APIs. But some of the win32 APIs which would make
this
easy are not available on XP, such as the InitOnce* group of
functions.
In the case of libav, it would also greatly simplify the threading code
thanks to native support for condition variables, in particular.
Windows XP is ancient, and despite its popularity is not even
supported
by Microsoft anymore. Windows 2003, which is on a similar level as XP
in supported APIs, is also being killed of by Microsoft. (Extended
support apparently ends in 13 days.)
So the question is: would it be ok to finally drop XP support for the
sake of cleaner and simpler code?
I would rather have phrased the question more explicitly as: "Is it
acceptable to drop 2003/older support in future versions?". Nobody
breaks support for already running software on existing computers
running XP in any case.
In my opinion, imposing known bugs on users of supported Windows
version for the sake of users of unsupported versions is immoral.
On the development side, the fact of the matter is that Microsoft
silently removes compatibility informations from MSDN for unsupported OS
versions. Indeed, Windows XP is arleady partly removed. So expecting
support in patches/reviews is impractical.
--
Rémi Denis-Courmont
http://www.remlab.net/
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
