This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH 05/10] Remove __need macros from errno.h (__need_Emath, __need_error_t).


On Thu, 11 May 2017, Andreas Schwab wrote:

> On Mai 11 2017, Joseph Myers <joseph@codesourcery.com> wrote:
> 
> > * Requiring GCC 5 as minimum GCC version for building glibc on the basis 
> > of what's maintained upstream (though that's more aggressive than we've 
> > previously been on increasing GCC version requirements).  (Versions before 
> > 4.9 don't work with build-many-glibcs.py anyway.)
> 
> openSUSE Leap 42.2 (aka SLE 12) has gcc 4.8 as the system compiler.

And what binutils version?  (Given the evidence that 2.24 is already too 
old.)

I think there are plausible arguments for requiring 4.8 with 4.9 required 
for glibc 2.27 (consistent with the times when we introduced 4.6 and 4.7 
requirements); 4.9 now (a bit more aggressive on requiring new versions, 
and minimum for use with build-many-glibcs.py; supports _Atomic and 
stdatomic.h, which would allow conform/ expectations to be added for 
stdatomic.h without XFAILs for them, even though we should avoid libatomic 
dependencies in glibc); 5 now (oldest version currently maintained).

I think there is a clear argument for requiring binutils 2.25 or later, 
given those x86 internal linker errors.

I think there is a clear argument for removing the NaCl port (and 
generally saying that while there isn't a particular order in which 
toolchain components have to go upstream, to accept a glibc port the GCC 
port should be upstream not long after the glibc port is in).

-- 
Joseph S. Myers
joseph@codesourcery.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]