Re: libtool-2.2.6a breaks build of guile-1.8.5, libtool-1.5.26 is OK

[Sergei Steshenko]

--- On Mon, 11/10/08, Bob Friesenhahn <address@hidden> wrote:

> From: Bob Friesenhahn <address@hidden>
> Subject: Re: libtool-2.2.6a breaks build of guile-1.8.5, libtool-1.5.26 is OK
> To: "Sergei Steshenko" <address@hidden>
> Cc: "Libtool Bugs List" <address@hidden>
> Date: Monday, November 10, 2008, 2:33 PM
> On Mon, 10 Nov 2008, Sergei Steshenko wrote:
> > 
> > Do I reed you correctly, i.e. can your statement be
> expressed this way:
> > 
> > guile can _not_ be built by libtool-2.2.6a out of the
> box using the
> > standard
> > 
> > ./configure
> > make
> 
> Normally upgrading libtool requires a somewhat
> package-dependent 'bootstrap' (libtoolize --force)
> procedure.  Considerable care is sometimes required to make
> sure that all residue is removed since libtoolize updates or
> overwrites and does not remove residue (to best of my
> knowledge).  Even if it did remove residue, the residue
> might be ressurected by the source control system.  This
> means that care needs to be taken than m4 macros from older
> libtool are not accidentally included with the newer libtool
> and legacy ltmain.sh scripts are removed.  Automake may
> become confused if it sees legacy files present.
> 
> Bob

I am at the moment is interested in a simple answer to the question whether
the discovered behavior guile-1.8.5 bug. I think it is.

I remember that other targets' build mechanisms are smart enough to check 
'libtol' availability at system level and if it is not available, to use
the built-in one.

With guile-1.8.5 it's strange/awkward - its 'configure' _demands_
'libtool'  at system level, but after detecting it uses the built-in one.

Which caused this problem to happen in the first place.

Regards,
  Sergei.