svn commit: r528258 - in head/emulators/virtualbox-ose: . files

[Kyle Evans]

On Sun, Mar 22, 2020 at 5:40 PM Gerald Pfeifer <gerald at pfeifer.com> wrote:
>
> On Thu, 12 Mar 2020, Kyle Evans wrote:
> > Log:
> >   emulators/virtualbox-ose: use contemporary GCC instead of old llvm
>
> This patch does not exactly do that. =:o
>
> > +# machine/atomic.h may use features that are only available in GCC9.  This is
> > +# a workaround, anyways- we should be expressing USES= compiler:c++14-lang, but
> > +# contemporary clang miscompiles virtualbox in some fashion and yields runtime
> > +# breakage.
> > +USE_GCC=     any
>
> USE_GCC=any allows using the age old, rotten, GCC 4.2 based system
> compiler where still present.
>
> This is practically not the case with ONLY_FOR_ARCHS= i386 amd64 which
> this port also specifies, but that is a subtle and indirect approach.
>
> (Where there is no /usr/bin/gcc USE_GCC=any degenerates to USE_GCC=yes,
> so the two are practically equivalent here, but still the USE_GCC=yes
> form is the one requiring a current version.)
>
> Okay to change this to USE_GCC=yes?
>

It's actually needing to be changed back away from GCC, because this
inherently breaks as the qt bits it uses are compiled with llvm and
used an incompatible libc++ -- there's an open PR about it after this
commit, I'm trying to work out where newer clang is failing to compile
properly. =-(

> Gerald
>
> PS: How do patches to this port actually get approved?  The log has
> hardly any Approved by-s.

I seem to recall the situation is that there's now an effectively
one-man vbox@ team, and this seems to be mostly on life-support. There
is a long-standing PR to push virtualbox-ose to 6.x, but I don't
recall if that was being ushered through by vbox@ or not.