This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: RFC: requiring GCC >= 4.7 to build glibc
- From: Joseph Myers <joseph at codesourcery dot com>
- To: David Miller <davem at davemloft dot net>
- Cc: <libc-alpha at sourceware dot org>
- Date: Mon, 5 Oct 2015 11:36:12 +0000
- Subject: Re: RFC: requiring GCC >= 4.7 to build glibc
- Authentication-results: sourceware.org; auth=none
- References: <alpine dot DEB dot 2 dot 10 dot 1508201344140 dot 30940 at digraph dot polyomino dot org dot uk> <alpine dot DEB dot 2 dot 10 dot 1508242014240 dot 23857 at digraph dot polyomino dot org dot uk> <20150824 dot 134349 dot 385142835363137131 dot davem at davemloft dot net> <alpine dot DEB dot 2 dot 10 dot 1508242224580 dot 23857 at digraph dot polyomino dot org dot uk>
On Mon, 24 Aug 2015, Joseph Myers wrote:
> On Mon, 24 Aug 2015, David Miller wrote:
>
> > From: Joseph Myers <joseph@codesourcery.com>
> > Date: Mon, 24 Aug 2015 20:28:34 +0000
> >
> > > Would anyone else like to comment on (a) the general principle of doing
> > > time-based upgrades of the minimum GCC and binutils versions for building
> > > glibc (so typically upgrade every other glibc release cycle, since GCC and
> > > binutils have major releases about once a year) or (b) this particular
> > > proposed increase?
> >
> > It can be a pain to upgrade gcc, and in fact for me this is much
> > harder than updating the kernel for example. If I upgrade gcc it
> > effects all of my development work, not just the work I do with glibc.
>
> What criteria would you like for the minimum GCC version for building
> glibc?
I haven't seen any responses to this question. Would those opposed to the
4.7 upgrade at this time in particular, or to my proposed principle of
time-based upgrades of the minimum GCC version in general, like to propose
alternative criteria? (E.g., "the current long-term-support releases of
GNU/Linux distributions X, Y and Z should include a suitable GCC version"
- or "support for release N can be dropped once release N+5 is out" (which
would imply dropping 4.6 support when GCC 6 is out).)
And, in light of
<https://sourceware.org/ml/libc-alpha/2015-09/msg00749.html>, what would
your views be on version upgrades for particular architectures when needed
to add new ABIs (__float128 support for powerpc64le and possibly other
hard-float powerpc variants), given that such version requirements might
be much stronger than a requirement for 4.7 (as in, no current released
GCC version yet has the __float128 support for powerpc)?
Note that I think a 4.7 requirement brings *more* benefits than most such
single-version increments of the minimum GCC version. And annual
increases in the required version do *not* require glibc developers to
upgrade the GCC version they use to build glibc annually, but only about
once every four years (if someone moves from 4.6 to 5.x now, they wouldn't
need to upgrade again until the required version reaches 6.x).
--
Joseph S. Myers
joseph@codesourcery.com