libtool-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SCO/bugfix patch 7 of 10: Improve SCO platform support


From: Ralf Wildenhues
Subject: Re: SCO/bugfix patch 7 of 10: Improve SCO platform support
Date: Wed, 9 Nov 2005 08:55:15 +0100
User-agent: Mutt/1.5.11

Hi Tim,

* Tim Rice wrote on Wed, Nov 09, 2005 at 06:57:16AM CET:
> On Tue, 1 Nov 2005, Ralf Wildenhues wrote:
> > * Kean Johnston wrote on Mon, Oct 31, 2005 at 06:43:03PM CET:
> > > >Patch 7 of 10 attached ...
> > > 
> > > Here's a re-submission using install_libdir instead of the
> > > 'instrpath' hack I invented.
> > 
> > Ugh.  It's really difficult to judge over this huge patch -- would have
> > been much easier without the SCOABSPATH part.  Also, I haven't checked
> > closely yet (other eyes appreciated!), so I'll only give a few
> > suggestions:

Well, I believe the SCOABSPATH is not only ugly, but also broken (from a
libtool perspective): if you have a package which creates two libraries,
one depending on the other, your uninstalled library will link against a
previous installed version, if that exists.  While testing, this is
wrong.

I like the idea of enabling absolute sonames though, but as a general
Libtool option:
- it's rather non-invasive to implement,
- it degrades gracefully: some systems (like AIX, I believe, or any
  static-only system) won't need it anyway, on other systems where it
  doesn't work, we can just not do it and be fine,
- it bears no special relation to UnixWare.

So, I ask you or Kean to resubmit the patch without the SCOABSPATH
parts, then I'll take another look again with the help of the manpages
you pointed to.

Also, I'll put adding above on the TODO list.  I'll not accept it before
2.0, but I'll be glad to assist in producing a patch against any current
version and/or test one, should you or Kean want or need this any earlier.

I believe it could work like this:
- We add an option `-absolute-soname' to the link-mode flags, and
  document it.
- We add a (non-tagged) _LT_DECL, maybe named allow_absolute_soname.
  It would be `no' by default, `needless' on AIX, `yes' on GNU/Linux,
  UnixWare, ...  (I'm not sure about the default, the encoding may be
  cheaper the other way round, given suitable catch-all values.).
- If the flag is given, and allow_absolute_soname=yes, and we are
  creating a library (not a program), we make sure relinking is needed,
  and mangle the soname appropriately (differently for the uninstalled
  vs. the installed library; or maybe only for the installed one, I'm
  not sure).
- We add a test to make sure this feature actually works.

It would then be easy to update `allow_absolute_soname' for other
systems as we go along.

I expect the ltmain changes to be pretty small, it's just the testing of
them on different systems that needs real work.

We can still see about whether there should also be a configure option
to enable this, or we should start reading options from an environment
variable, to facilitate building a whole bunch of packages with this
flag enabled.  This is an (or two) orthogonal issue(s), though.


If you agree with me on this plan, then all is well.  If you disagree,
I'd like to hear arguments.  My mind is not set in stone, neither am I
the only one to decide.

With above, you can still have your patches against your local
Libtool-1.5 if you're really dying to; both profit: we have less cruft
and more general features, and you have your feature, and with time, you
won't need a patches Libtool any more.  Surely it'll be more work, and I
just don't have the time to do the patch myself right now.

> I don't know if this is any help or not, but here are the
> before and after stats on the (sco) platforms I have running.
> http://www.multitalents.net/branch-1-5-status.html

[ corrected the obvious typo ]

Thanks.  It would be even better if you could show the output of the
failed test groups with VERBOSE=x set.  :)
(remember you have to run the tests in groups for them to work)

There are a few new failures.  I wonder why they fail now but did not
before the patch.  Maybe the verbose output of the ones that passed
would help, too.

> > By the way, is there good online documentation for these systems' ld and
> > dynamic loader?
> 
> ld man pages are here.
*snip*

Thank you very much!

Cheers,
Ralf




reply via email to

[Prev in Thread] Current Thread [Next in Thread]