Major CVS LFS bug - binutils-2.13.2.1

Dagmar d'Surreal dagmar.wants at nospam.com
Mon Jan 27 23:07:15 PST 2003


On Mon, 2003-01-27 at 20:56, Gerard Beekmans wrote:
> On January 27, 2003 07:51 pm, Ian Molton wrote:
> > it appears to have a failed test. surely a workaround is in order, like
> > our other packages?
> 
> Sure a hack can be put into place as always.
> 
> But more often we've seen that the HJ Lu releases fix bugs a lot quicker. It's 
> even the preferred binutils release by the kernel developers. That ought to 
> mean something too.
> 
> It's not really jumping ship in the bad sense. It's just that every single 
> time we find a problem with Binutils, it appears that the problem does not 
> occur in the HJ release. This might lead one to believe that using HJ's 
> release will prevent a host of future problems that might arise by using the 
> standard GNU releases.

Point of clarification: there is no problem occurring in the GNU
binutils release this time.  It occurs in gcc and glibc.  Dropping back
to 2.13.2 seems a perfectly reasonable kludge to me, considering that
fixing the scripts in gcc and glibc is likely to cost someone a sanity
point.  This "every time" business is a little misleading.

> > also, HJL binutils is NOT cross platform, the book is.
> 
> I thought it was. True, x86 gets most fixes but I didn't think the other 
> platforms were removed from it. Correct me if I'm wrong though.

I dunno about HJLs releases not being cross-platform, since he
originates fixes for a few platforms other than just x86 (he's a mac
daddy of assembly), but the fact that these fixes have only been tested
by one guy at the time of release should be cause for at least mild
concern.

I suggest a two-tiered plan of attack to save wear & tear on end-users. 
For _now_, since we know that it won't cause breakage for binaries drop
back to 2.13.2.  In the space of say, a week or two, once we're more
certain there's nothing ugly and scary in the _beta_ release HJL emits,
then change the book to use the HJL version in the interim while we wait
for the next boringly-stable and cross-platformly tested GNU binutils
with a less lengthy version number.

-- 
The email address above is just as phony as it looks, and for obvious reasons.
Instant messaging contact nfo: AIM: evilDagmar  Jabber: evilDagmar at jabber.org

-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message



More information about the lfs-dev mailing list