Re: devel/libffi on ARM

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Wed, Nov 12, 2014 at 08:46:07AM +0900, OBATA Akio wrote:
> On Wed, 12 Nov 2014 00:57:58 +0900, Joerg Sonnenberger <joerg%britannica.bec.de@localhost> wrote:
> 
> >On Tue, Nov 11, 2014 at 10:10:34AM +0900, OBATA Akio wrote:
> >>On Tue, 11 Nov 2014 01:56:09 +0900, Joerg Sonnenberger <joerg%britannica.bec.de@localhost> wrote:
> >>
> >>>On Mon, Nov 10, 2014 at 03:44:42PM +0000, John Klos wrote:
> >>>>It seems that devel/libffi/patches/patch-src_arm_sysv.S causes
> >>>>libffi to fail on NetBSD 7 on evbarm:
> >>>
> >>>Correct, the person doing the update broke it. Just dropping the patch
> >>>is not correct as it will result in missing unwind annotations.
> >>
> >>Send patches to upstream first, or stay on old obsoleted version forever?
> >
> >Or maybe just don't do drive-by updates without understanding all
> >patches OR testing them?
> 
> Then, will I send all future update request to you after if you applied local patches, OK?

I don't remember you updating core dependencies (i.e. Python depends on
libffi), but sure -- if there is a decent enough chance of breakage and
a conflict is a good indicator...

> Or have you tested on ARM platforms other than NetBSD/eARM with various compilers/assemblers?

I know that it works on all ELF platforms with non-ancient binutils. It
likely works on iOS too, but feedback for that is welcome. Noone
complained about the status before the update.

Consider some common sense -- packages with a known history of breaking
things on update (Boost anyone?) or a high degree of platform specific
logic (compilers, libffi, most JIT implementation) should be updated
with care. "Compiles on platform X" alone is not a good indicator of
that.

Joerg