This is the mail archive of the
mailing list for the glibc project.
Re: Require GCC 6.2 or later to build glibc
- From: Joseph Myers <joseph at codesourcery dot com>
- To: "Gabriel F. T. Gomes" <gabriel at inconstante dot eti dot br>
- Cc: <libc-alpha at sourceware dot org>
- Date: Fri, 1 Feb 2019 16:31:55 +0000
- Subject: Re: Require GCC 6.2 or later to build glibc
- References: <alpine.DEB.firstname.lastname@example.org> <20190201120302.GA5116@tereshkova.br.ibm.com>
On Fri, 1 Feb 2019, Gabriel F. T. Gomes wrote:
> On Thu, Jan 31 2019, Joseph Myers wrote:
> > Thus, this patch increases the minimum to 6.2, removing the
> > documentation of the separate requirement for powerpc64le. It does
> > not remove the powerpc64le configure test, or any __GNUC_PREREQ that
> > could be removed as not being in installed headers or files shared
> > with gnulib; I think such cleanups are best done separately.
> I'd like to keep the tests for powerpc64le requirements, namely, the
> checks for the presence of quad-precision builtins, and POWER8 as target
> (sysdeps/powerpc/powerpc64/le/configure.ac). Was it your intention to
> suggest that we should remove the check for the builtins or are these
> cleanups about __GNUC_PREREQ alone?
I'd suggest removing the libc_cv_compiler_powerpc64le_binary128_ok
configure test, as well as __GNUC_PREREQ calls, since the minimum GCC
version should imply those built-in functions are available. I think the
point of such configure tests (for conditions that, if not met, would
result in a build failure anyway) is to ensure a helpful early error
saying what the problem is in the most likely cases of something wrong
with the build tools (such as GCC being too old) - not to cover every
theoretically possible form of compiler breakage (such as a 6.2 compiler
somehow having had support for those built-in functions removed).
Joseph S. Myers