Re: RFC: requiring GCC >= 4.7 to build glibc

[Zack Weinberg <zackw at panix dot com>]

On Fri, Aug 21, 2015 at 11:23 AM, Mike Frysinger <vapier@gentoo.org> wrote:
> On 20 Aug 2015 16:36, Joseph Myers wrote:
>> On Thu, 20 Aug 2015, Mike Frysinger wrote:
>>> On 20 Aug 2015 13:51, Joseph Myers wrote:
>>>> * More C11 support, so we no longer need to XFAIL some conform/ tests and
>>>> could use the new features in glibc if desired.
>>>
>>> shouldn't we already be handling this and not XFAIL-ing when the compiler
>>> is new enough ?
>>
>> Although such configure tests would be possible, they'd seem like
>> overkill.
>
> i'm not suggesting feature tests.  we're already testing & have access to the
> compiler version which means it should be possible to leverage that data.

I looked into this exact thing myself, last week, because the conform/
XPASSes were annoying me.

The configure script currently doesn't expose the GCC version number
to the Makefiles, which wouldn't be hard to change, but what *would*
be a pain is getting it into the Makefiles in a usable form.The
problem is that, as far as I can tell, the only way to do a
conditional XFAIL is with if-statements in the relevant Makefile ...
and gmake conditionals can't do numeric comparisons.  (In fact, as far
as I can tell, there is no way to do arithmetic of any kind in the
gmake language, short of $(shell ) or $(guile ), the former being slow
and awkward, and the latter not necessarily available.)  So you would
need a whole series of substitution variables,
GCC_VERSION_AT_LEAST_4_6 through
GCC_VERSION_AT_LEAST_some_upper_limit, and frankly that seems like
more trouble than it's worth, if all we're getting out of it is the
elimination of three XPASSes.

zw