[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: icc/ecc full path and solaris link line fixes
From: |
Ralf Wildenhues |
Subject: |
Re: icc/ecc full path and solaris link line fixes |
Date: |
Wed, 17 Nov 2004 17:34:24 +0100 |
User-agent: |
Mutt/1.4.1i |
* Mats Rynge wrote on Wed, Nov 17, 2004 at 04:34:52PM CET:
> On Wednesday 17 November 2004 06:03 am, Ralf Wildenhues wrote:
> >
> > > Looks good. Since I don't have access to Solaris: Is there a
> > > testcase in the testsuite that is fixed by this? If not, we should
> > > maybe create one. It'd also be great to know how the branch-2-0
> > > testsuite fares on Solaris, anyway.
> >
> > Are you still reading here? Can you or someone else confirm that
> > demo-nopic.test fails on Solaris? It's mentioned in
> > http://lists.gnu.org/archive/html/bug-libtool/2004-09/msg00171.html
> > and I'd disable that test for the system then.
>
> Yeah, sorry for the delay. I still haven't looked to see if the
> extractall is covered by a unit test, but you are right about the
> demo-nopic.test:
Thanks for checking!
Now we have two choices (w.r.t the message I quoted above):
Enable -mimpure-text on sparc-solaris/gcc which would allow non-PIC code
in dso's. Or disallow non-PIC in dso's (and adjust demo-nopic.test).
Or create a dedicated flag (let's call it `impure_flag') for non-PIC
code in dso's.
I vote for the second option, since portable code should not depend on
it (in the name of a sensible common subset of features). What do the
others think?
Second, I think Ralf Menzel is right that Solaris wants
- _LT_AC_TAGVAR(no_undefined_flag, $1)=' -z text'
+ _LT_AC_TAGVAR(no_undefined_flag, $1)=' -z defs'
Do the others agree?
Regards,
Ralf