On Sat, Feb 18, 2012 at 10:37:14PM +0800, Oleg Tsarev wrote:
> 2012/2/18 Mark de Wever <[email protected]>:
> > On Sat, Feb 18, 2012 at 11:52:53AM +0800, Oleg Tsarev wrote:
> >> Hi all,
> > At the moment I compile test:
> > - Debian stable
> >  - gcc 4.3 in C++98 mode
> >  - gcc 4.4 in C++98 mode
> >
> > - Debian unstable in a chroot
> >  - in C++98 mode
> >    - gcc 4.4
> >    - gcc 4.5
> >    - gcc 4.6
> >    - gcc 4.7 (from experimental)
> >    - gcc snapshot
> >    - clang 3.0
> >  - in C++11 mode
> >    - gcc 4.5
> >    - gcc 4.6
> >    - gcc 4.7 (from experimental)
> >    - gcc snapshot
> >    - clang 3.0
> >
> 
> 1) I can use different gcc
> 2) I can setup "sequentianal" build (run first build with gcc X, than
> second with gcc Y)

I do the same.

> 3) I use OpenVZ - this is advanced chroot (by design).

I tried virtualization with Virtualbox, but the I felt the chroot work
better for me. Since you're the one who's going to implement it, you
should pick what you think works best for you.

> I am going to support all compilers which you listed -

I'm not entirely sure whether this is a good idea. The gcc-4.7 and
gcc-snapshot are experimental compilers. Gcc adds new default warnings
with newer compiler versions and since warnings are errors with our
default switches the compilation will fail. This mean the CI tests fail
after a random compiler update (especially gcc-snapshot; it's a gcc
development snapshot). I rather see the CI tests use released compilers
instead of their development snapshots.

I'm also not sure whether you want to run the C++11 mode tests, but I
also think that's not a good idea. Our tree is C++98 and I don't expect
that to change for the near future, since compiler support is not good
enough yet. Some perfectly legal C++98 code is invalid in C++11, for
example an SDL_Rect initialised with brace initialisation is no longer
allowed to narrow in C++11.

So I propose not to run those compile tests, or if they break regard
them as non-important breakage. It still should be fixed of course, but
too much unexpected breakage might discourage people to pay attention to
the tests. I still want to keep my chroot script running for these
compilers and C++11 mode to make sure once we switch to a newer gcc or
C++11 mode the switch will be smooth.

I look forward to see the CI up and running.

-- 
Regards,
Mark de Wever aka Mordante/SkeletonCrew

_______________________________________________
Wesnoth-dev mailing list
[email protected]
https://mail.gna.org/listinfo/wesnoth-dev

Reply via email to