Re: Mac OS X 10.4 test failures on HEAD

[Ralf Wildenhues]

Hi Peter,

* Peter O'Gorman wrote on Fri, Jun 17, 2005 at 04:29:57PM CEST:
> Ralf Wildenhues wrote:
> >>In case anyone is wondering (Gary) why HEAD fails one of the new tests on 
> >>Mac OS X 10.4, it is because of a bug in the linker. I have reported the 
> >>bug and it is unlikely to be fixed before Mac OS X 10.5, I think.
> >
> >
> >Which one fails, and how?  Just so we can distinguish it easily from
> >genuine failures.
> 
> Hi Ralf, sorry, had to rerun the testsuite because my memory was not good 
> enough to remember "template test with subdirs", failure is here:

One nice thing about Autotest is that it helps exactly with this.

Which failed? Either
$ less testsuite.log

or
$ ls testsuite.dir

tells you (the directories of the failed tests are still present).  Then,

$ less testsuite.dir/NUMBER/testsuite.log

has the error again, and all the other files to reproduce it should be
present, too.  (Libtool's suite might need tweaking for this to work
correctly.)

*snip bug output*
> Note that this looks like a bug in the dynamic linker (dyld) it is not, the 
> bug is in ld.

OK.

> >>There is a pretty good workaround though, set the 
> >>MACOSX_DEPLOYMENT_TARGET environment variable to 10.3 or later and 
> >>./configure; make; make check will pass all tests.
> >
> >Thanks for this information.  Should this be in doc/notes.texi?
> 
> Dunno, what do you think?

I'm kind of unsure whether Libtool would be the right place for a bunch
of linker-specific curiosities similarly to Autoconf's `Limitations of
Usual Tools'.  In any way, it might not hurt to record the information
somewhere (and, what's more, your cut'n'pasting of the error will allow
this to be picked up by search engines).

Regards,
Ralf