Re: gcc-4.6

[Thorsten Glaser <tg@mirbsd.de>]

Finn Thain dixit:

>On Fri, 28 Oct 2011, Thorsten Glaser wrote:
>
>> Also, did anyone _test_ it, beyond what a bootstrap does?
>
>A year ago I used the 4.6 branch in the form of a cross compiler (C and
>C++, no GCJ). I didn't bootstrap it and I didn't test it much.

Mh. I meant the packages uploaded to Debian unstable, though.

>Seems to me that if some portion of the archive can't be built and if

I think the archive _can_ now be built. I’m trying to set up an
ARAnyM buildd (like the zlin* machines used to be), and others
are (hopefully still) trying to resurrect the real iron ones.
It’s just a matter of time.

>switching compilers won't significantly reduce that portion, then that
>time and effort may be better spent on other issues.

I think it’s the other way round – keeping with gcc-4.4 gets us
problems (I’ve seen some) because everyone else, except Alpha,
uses gcc-4.6 by now. (And possibly hppa.)

On the other hand, gcc-4.4 with all the current patches applied
_has_ proven solid. The kernel builds also still use it, but to
switch the default userspace compiler might be possible (pending
the gcj issue, Debian #633754, PR49847).

>IMHO changing the default compiler should be deferred until gcc 4.4
>becomes a problem.

I don’t really have an opinion either way, I just wanted to point
out that the porter team should make a decision (which of course
wouldn’t imply it can’t be changed), also in the light of the up-
coming gnat-4.6 patch. I just don’t want m68k to start lagging
behind again, after all the work. After all, I won’t stay with
this port forever.

bye,
//mirabilos
-- 
I believe no one can invent an algorithm. One just happens to hit upon it
when God enlightens him. Or only God invents algorithms, we merely copy them.
If you don't believe in God, just consider God as Nature if you won't deny
existence.		-- Coywolf Qi Hunt