svn commit: r316102 - in head/emulators/dolphin-emu-devel: . files

[Chris Rees]

On 21 Apr 2013 15:45, "Alexey Dokuchaev" <danfe at freebsd.org> wrote:
>
> On Sat, Apr 20, 2013 at 01:48:52PM +0000, Ganael LAPLANCHE wrote:
> > Then, I see three possibilities :
> >
> > 1) Fix the BROKEN message with something more incentive, like just :
> >
> > .if ${ARCH} == "powerpc"
> > BROKEN= Not tested on powerpc
> > .endif
>
> Bad idea.  If port breaks during the build (that's when we mark it as
> BROKEN), just state the reason (try to be concise yet precise), and maybe
> quote relevant part of the build log in commit message (just a few lines
> would suffice).
>
> BROKEN has nothing to do with the fact if software in question was tested
> or not, whatever "tested" might mean.
>
> > but I don't find this solution very clean as, after all, I have not
> > tested other tier-2 architectures, so why just put a message for
powerpc ?
>
> Nor is it relevant if was was tested on any tier-X arch.  BROKEN is used
> when something does not build (or, more strictly, does not build on the
> cluster).
>
> > 2) I could rewrite it this way :
> >
> > .if ${ARCH} != "i386" && ${ARCH} != "amd64"
> > BROKEN= Only tested on i386 and amd64
> > .endif
>
> Just as bad, for the same reasons.
>
> > 3) As I don't know exactly why this BROKEN message has been introduced
> > and whether the port now builds or not on powerpc and other tier-2 archs
> > (I don't have hardware to test it),
>
> Unfortunately, not just you, most of us have no access to tier-2 hardware.
> However, we need to generate packages for those releases, so naturally, we
> have such boxes in the build cluster, and portmgr@ guys will occasionally
> mark some ports as BROKEN on them.  Sadly, most of the time we cannot do
> anything about it (except for few trivial cases of when pluto/flame is
> enough to reproduce the error).
>
> I would probably try to locate the recent powerpc build log, see why it
> fails, and provide better BROKEN message.  If the port builds fine now
> (there is a slight chance), just remove it for good's sake.  ;-)

Mark may correct me here, but I seem to remember that if there's a chance
of the change fixing it you should just remove the tag, and a portmgr will
readd if necessary.

By the way, you should not quote strings in Makefiles.

Chris