Re: GCC version change / C++ ABI change
On Mon, Jul 04, 2005 at 11:44:23AM +0200, Thiemo Seufer wrote:
> Horms wrote:
> > On Mon, Jul 04, 2005 at 11:12:21AM +0200, Thiemo Seufer wrote:
> > > Otavio Salvador wrote:
> > > > Thiemo Seufer <ths@networkno.de> writes:
> > > >
> > > > > Junichi Uekawa wrote:
> > > > >> Hi,
> > > > >>
> > > > >> > This week, we will change the GCC default versions from 3.3 to 4.0
> > > > >>
> > > > >> Would it break kernel 2.4 builds somehow ?
> > > > >> I've not been quite following; but the thread almost a month ago
> > > > >> seems to indicate thus:
> > > > >> http://www.kerneltraffic.org/kernel-traffic/kt20050701_316.html#7
> > > > >
> > > > > Quite likely, yes. 2.4 Kernels would need to Build-Dep on 3.4.
> > > >
> > > > But the current versions of 2.4 doesn't get fixed yet?
> > >
> > > Most kernel hackers don't care that much about 2.4 any more.
> >
> > I'd rephrase that as, we need to discuss if 2.4 should be included
> > in etch.
>
> I don't think gcc-4.0 is a hard requirement for that. We still have
> even gcc-2.95 in the archive, and a gcc 3.3/3.4 version is likely to
> be around for etch.
Sure. I don't think there is any immediate threat that we won't be able
to compile 2.4 in etch. For i386 at least, we have been using gcc-3.3 to
compile 2.4.27, and from a casual glance that seems to be the case in
other arches. I am not sure about 3.4's ability to compile 2.4.27, but
it seems unlikely to me that all of the gcc versions you mention above
will be omitted from etch.
> > My understanding is that it is needed for some arches,
> > and my personal feeling is that 2.4 is maintained upstream and in
> > many cases is a valid choice over 2.6.
>
> I just wanted to hint that upstream is more interested in making 2.6 a
> more valid choice instead of sinking time in a compiler upgrade for 2.4
> which provides little benefit for the kernel.
Yes. I understand from a recent lkml post that there is
absolutely no interest in making 2.4 compile with gcc 4.
And yes, discussing 2.4's incusion (or not) in etch isn't directly
related to this discussion about gcc versions at all. I just wanted
to bring it to the table so people can mull over it a bit.
Though I might have been better off making a separate post.
--
Horms
Reply to: