Bug reports for the GNU libtool shared library maintenance tool

Ingo Krabbe | 2 Oct 2010 10:41

Re: break up functions

> 
> We have been taking some smaller steps here in the past, and I think
> that is a much more viable way than a massive breakup.  But anyway I'll
> be approving patches here only from contributors who then have the
> resources to fix regressions in the months afterwards.

Hi Ralf, 

actually I would take care of libtool a bit more, as I always ignored it
in the past, since it wasn't ever doing the things I wanted from it and
I never got a step into the code.

Now, it seems though, I found my way into reading that stuff.  If we are
talking about "months", and it will be ok, to have there one or two
months in between where I couldn't fix something, I would actually be
able to take care about libtool development and "fix regressions".  When
pointed to such bugs, that should be quite easy.

... and for the rest of the list readers ...

But still, what bothers me more, is what do you think about the
library/wrapper issue?  Do you see this problem, as I do, to be
critical?  And what should be the steps to get it corrected?

bye ingo

headers

Ralf Wildenhues | 2 Oct 2010 15:02

Re: break up functions

* Ingo Krabbe wrote on Sat, Oct 02, 2010 at 10:41:14AM CEST:
> Now, it seems though, I found my way into reading that stuff.  If we are
> talking about "months", and it will be ok, to have there one or two
> months in between where I couldn't fix something, I would actually be
> able to take care about libtool development and "fix regressions".  When
> pointed to such bugs, that should be quite easy.

That's one part of the problem: the initial bug analysis (until we know
what caused the bug) is what often takes most of the time.

> But still, what bothers me more, is what do you think about the
> library/wrapper issue?

It looks like a bug.

> Do you see this problem, as I do, to be critical?

Not sure what you mean here.  FWIW, I haven't encountered the bug in
practice.  Generally speaking, only regressions and bugs that are widely
observed are critical enough to hold up a release, and the latter only
if we have a patch, or can produce one in a reasonable amount of time.

Other than that, we sure would like to fix all bugs, but there is only a
finite amount of resources.

> And what should be the steps to get it corrected?

First step should be a simple and reliable reproducer, ideally in the
form of a testsuite patch.

(Continue reading)