[Groff] Pre-commit testing, automake

[Peter Schaffter]

On Sun, Jun 22, 2014, Ingo Schwarze wrote:
> However, my impression is that at least part of the problem we are
> facing here, and probably the more important part, is social in
> nature rather than technical.

Change is in the air.  It'll be a while before we adapt to it
fully.

During Werner's tenure, we had relatively few contributors and God
at the helm.  After the spate of discussions about groff's future,
etc., we're beginning to attract fresh blood.  Our situation now is
that we have more contributors, or potential contributors, but no
longer have the luxury of trusting just one person to spot problems.

For the next while, it's important that we go overboard in testing
our work before committing--the social aspect Ingo's talking about.
It's a pain, I know, submitting patches to the list, but, for now,
it's as good a way as any to deal with what Vaibhaw points out:

  "Groff seems to be complex enough for not just one person to get
   their heads around."

Vaibhaw has intimated he will attack the issue of "test suites
around major packages that can quickly sanitize our checkins or
an automated build and test system", and Betrand has submitted a
proposal for migrating to automake.  My feeling is that both are
important (automake perhaps a little less so, see below) now that
there are more people wanting to contribute to the project.  Groff
has been a pretty closed community for the past decade so we haven't
had to deal these things; now we do.  During this what amounts to
transitional period, extra vigilance with respect to changes and
commits needs to be practised.

On the automake debate, I favour migration but have no strong
opinions.  I know others do, and I'm wondering if those with
objections could post them for discussion so Betrand's work won't
be in vain should some compelling reason for leaving things
as they are emerge.  Vigilance, again. :)

-- 
Peter Schaffter
http://www.schaffter.ca