Time to obsolete arm-symbian?

[Alan Modra]

On Mon, Aug 17, 2020 at 09:07:04PM +0200, John Paul Adrian Glaubitz wrote:
> Hi!
> 
> On 8/17/20 8:34 PM, Jessica Clarke wrote:
> > As one of the porters in Debian (Adrian, the other, Cc'ed) I have some
> > concerns about this decision. As far as I can tell, the proposal was
> > made and a decision reached in just a single day? Moreover, this was
> > done hidden inside a thread about arm-symbian that no ia64 porter in
> > their right mind would notice and care about? Please, consider
> > reverting this. The Debian ia64 port is alive and happily using new
> > binutils versions as they get packaged for Debian, and if there are
> > bugs that actually matter we will find and fix them. The segfaults are
> > maybe an issue, but as far as I know we have not observed them
> > happening in the real world, so we are not aware of them nor do they
> > seem like a priority to address. Sergei Trofimovich (also Cc'ed) is a
> > Gentoo dev who is maintaining their ia64 port and fixed a GDB build
> > regression just recently. Both distributions have active users (and I
> > don't just mean ourselves). So, please, could you try and actually
> > involve your community before making instant decisions that have
> > serious consequences for them?

There are no serious consequences yet.  ia64 support is still in
binutils, but as of the patch I committed, needs --enable-obsolete at
configure time.  Assuming no one cares enough about ia64 to contribute
fixes for the segfaults, ia64 would remain in binutils until after the
next release, at which point the ia64 support code might be removed.

Incidentally the ld segfaults have been happening since at least
2.31, probably earlier.

> The irony here is that Anatoly (CC'ed) is currently setting up a new ia64
> porterbox for the GCC compile farm to provide a machine for testing :-).

-- 
Alan Modra
Australia Development Lab, IBM