Re: libtool pre-1.5b tests fail on 9 debian arches

[Bob Friesenhahn]

On Sat, 27 Sep 2003, Dalibor Topic wrote:

> I appreciate the hard work you're doing on keeping libtool in shape on
> debian. I can feel with you a little bit, as I'm hacking on kaffe's
> build system, and kaffe (in theory, at least) builds on 50+ platforms, a
> few of which I ocassionally get to test, and fix/let people fix libtool
> on. A few months ago I started feeding back kaffe's libtool patches into
> the main tree. It was hard work: it took about 2 months to get all of
> the 5 patches in, of which a few only changed a single line, almost all
> required a 'ping! single line patch uncommented for 2 weeks, does anyone
> maintain libtool?' mails and so on. I was angry at times, so angry that
> I considered forking libtool myself.

As I recall, when your repeated postings to the libtool list caused me
to feel guilty so I allocated some time to apply outstanding patches,
I discovered there were just a few outstanding usable patches spanning
a couple of weeks.  The situation was made out to be much more dire
than it really was.

In order for a patch to be applied, it needs a ChangeLog entry and the
actual author of the patch must be known (for copyright reasons).  If
the patch is large (15+ significant lines), then there must be a
signed agreement between the actual patch author and the FSF (also for
copyright reasons) and this paperwork takes time.  Patches which
change libtool flags or add features because "it worked for me" on one
system, but have unknown (or known unpleasant) consequences on other
systems are not likely to be accepted as is.

That said, I do know that there are currently several patches
outstanding (Charles Wilson's Cygwin patch, and Albert Chin-A-Young's
libtool tags patch) which still need to be applied.

Bob
======================================
Bob Friesenhahn
address@hidden
http://www.simplesystems.org/users/bfriesen