On 2015-01-05 3:58 PM, Philip Chee wrote:
On 05/01/2015 07:43, Mike Hommey wrote:
On Sun, Jan 04, 2015 at 02:28:30PM +0800, Philip Chee wrote:
To me, the default answer to whether we should keep supporting MinGW
is "no", merely because it will require time and effort that will not
directly benefit our users as we do not use that compiler to release
Firefox. That is, without someone coming up with a good reason
otherwise, we should drop it. And not having it locally installed is
not a good reason. :-)
There is a big difference, though. There is a benefit from compiling
with mingw, in that it's a free/libre toolchain, which MSVC isn't.
People who do want to build Firefox or a Gecko-based product can do so
without using a proprietary compiler. This has value to a lot of people.
But once you have to choose between one version of a proprietary
compiler and another, there is no such difference anymore.
How close are we to being able to compile Firefox with clang on Windows?
IIRC clang is free/libre - but not copy-left.
We can build Firefox with clang-cl on Windows right now but that still
depends on the rest of the Microsoft toolchain for now.
Note that we regularly break mingw builds, probably much more often than
we've broken MSVC 2012 builds, and we still accept fixes for those
breakages. The situation is no different with MSVC 2012.
Now, the question whether it's worth bothering with MSVC 2012 fixes is
an interesting one, because of that lack of philosophical difference
between 2012 and 2013. If you've come your way to install 2012, you can
just as well upgrade to 2013. The benefit is better support for modern
C++. And that, combined with the fact that 2012 has never been a
toolchain we use on automation, make a case for dropping support for
2012.
Looking at it another way, the only reason I can see that people are
currently using 2012 is that they wanted something more modern than 2010
for some reason (I guess there are IDE changes that are worth?). They've
had to endure build failures from time to time because 2012 was not
what's used on automation. I don't think anyone in their right mind
would have installed 2012 after we dropped support for 2010, because the
current version was 2013 at the time, and what's the point to upgrade if
it's not for the current version?
So, keeping support for 2012 is essentially keeping support for people
that decided to upgrade before everyone for some reason. They've had a
long run, they can be forced to upgrade now.
You are probably right but then it would have been better to have
unsupported VS2012 at the same time as VS2010 instead of dropping VS2012
only a brief interval after VS2010.
Yes, but we cannot change the past, and it's only been three weeks or so
since we dropped VS2010.
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform
