Re: glibc 2.3.2 intermediate status of the compilation (not test) on all archs
At Thu, 10 Jul 2003 08:06:09 +0200,
Matthias Klose wrote:
>
> James Troup writes:
> > GOTO Masanori <gotom@debian.or.jp> writes:
> > > Matthias, could you tell me that --enable-sjlj-exception is really
> > > needed for some archs?
>
> arm still uses sjlj based exceptions in 3.3, so there is at least one
> arch which does not have a choice.
OK... Do you know any plans to change it?
> > Yes, it's needed. If we don't use it, we break binary compatability
> > on those architectures and that is simply not an option. (AIUI, FWIW
> > etc.)
>
> these are m68k and hppa. not sure if it's an option or not. at least
> for binutils glibc-2.3.1 shows failures where I see no ones with
> 2.2.3, is this fixed for 2.3.2?
I don't know because working m68k and hppa are not available.
> btw, would it help to have two different glibc versions, i.e. 2.3.1
> for arm, hppa and m68k and the new one for others?
Yes, it's one answer... glibc package can be splited up to
glibc-2.3.1 and glibc-2.3.2. However, I would like to make all
architecture use same glibc version as far as we can. Because newer
glibc has newer function (versioning, new standard, new kernel, and so
on). In addition, we don't want to be burried with the patched glibc
2.3.1 :-)
But if this situation continues for two or three months, I woule like
to choice your opinion.
Regards,
-- goto
Reply to: