Priorities in Automake developement (was: [PATCH] Modernize, improve and/or extend test scripts `conf*.test')

[Stefano Lattarini]

At Sunday 08 August 2010, Ralf Wildenhues wrote:
> * Stefano Lattarini wrote on Sun, Aug 08, 2010 at 03:11:13PM CEST:
> > At Sunday 08 August 2010, Ralf Wildenhues wrote:
> > > and I'm sorry there is such a high review delay.
> > 
> > Well, I assume this is not done on purpose (right? ;-), so don't
> > worry.
> 
> That depends on how you interpret it, I guess.  Testsuite work
> doesn't add new features to Automake.  We have only finite amount
> of time for Free Software.  If we spend all of that fixing the
> testsuite, there will be no time left for new features or bugs
> outside of the testsuite. And yes, with "we" I mean your time for
> Free Software as well.  :-)

All true; I'd just like to clarify some points and opinions:

 1. For a "newbie", hacking/analizyng the testsuite is a really good
    way to get accustomed to the overall functionalities of a software
    package, which is IMO a prerequisite for modifying the software
    itself (this is especially true if the testsuite has a low entry
    barrier, like Automake's one does).  So far, my experience with
    Automake seems to confirm this idea.

 2. Automake, in its current incarnation, is mostly feature-complete
    IMHO (it still lacks a way to easily write non-recursive makefiles,
    true, but it seems we are getting there finally).  What it's really
    needed is to make Automake *easily extensible* by third-parties,
    like e.g. Autoconf is.  Doing this, however, might require a heavy
    refactoring/reorganization of the codebase, and for this a strong,
    reliable testsuite is required.  Well, that, and a lot of time, and
    lot of community partecipation, and... you got the gist.

So I think that my time hacking the Automake testsuite is not badly
employed, after all.  And it's kinda fun, too :-)

Regards,
   Stefano