On Sat, 2005-07-09 at 13:27 +0100, David Woodhouse wrote:
On Fri, 2005-07-08 at 13:03 -0700, Ulrich Drepper wrote:
> Ralf Ertzinger wrote:
> > So glibc will have to be excluded from the update frenzy :)
>
> As I said, in a couple of days we'll hopefully have the gcc, binutils,
> glibc pieces in place to make the next big change in the build
> procedures and then you cannot do *any* update unless you get the new
> glibc. The rpm infrastructure should catch those requirements but be
> warned. We'll announce the build changes ahead of time as well.
Can we drop -fsigned-char on PPC at the same time?
Currently, it's massively inconsistent because some packages use
$RPM_OPT_FLAGS while others don't. If you build an executable from a C
file with just 'make testprog', you'll end up with different signedness
of 'char' from that which glibc was built with, etc.
My understanding was that glibc doesn't care which signedness char has.
(Although, other libraries may matter. Perhaps a section on not using
char in library ABIs is in order for dsohowto.pdf)
--
Nicholas Miell <nmiell(a)comcast.net>