Re: IPC Functions

[forkandwait]

Jordi Gutiérrez Hermoso <jordigh <at> gmail.com> writes:

> 
> 2010/8/23 Judd Storrs <jstorrs <at> gmail.com>:
> > 2010/8/23 Jordi Gutiérrez Hermoso <jordigh <at> gmail.com>
> >>
> >> Same as everything else, really. Are you going to be the one to
> >> maintain them? Are they currently in a good enough state to be in core
> >> Octave? Are there any licensing issues we need to take into
> >> consideration?
> >
> > If octave doesn't want them in the core, they shouldn't be listed in the
> > PROJECTS
> 
> I'm sure they should be there, and if it were up to me, all of 'Forge
> would be shipped along with Octave, as most people already use it that
> way, in accordance to what the competition does. However, there are
> are practical considerations and obstacles to take into account.
> 
> > file so that well-meaning people looking for some way to contriute
> > aren't immediately set upon by this snotty, bellicose attitude.
> 
> I didn't mean to be bellicose, and perhaps I'm unavoidably snotty
> sometimes, but it's a real concern and what I understand is the usual
> response to why more 'Forge functions aren't moved into Octave.

I think part of the problem is that we don't have a policy regarding extensions
to Octave past the ML core.  We could say "we mirror TMW exactly, including
toolboxes", but we don't, in the sense that no one is policing whether packages
like Optim follow TMW toolboxes.  We *mostly* keep Octave core a one-one mirror
of TMW, but even then we fudge and let non-standard functions like table() into
the primary codebase (a mistake in this case, as accumarray() does it better and
more standardly).  This lack of policy means that there is no clear line about
what to do with functions that aren't in the alphabetical list at TMW but which
are obviously useful; the easy punt is to say "throw it on the 'Forge".

We could start trying to hit conformity with the toolboxes (and bring them into
the main tree) -- is there an IPC toolbox? 

If we did that, I would also suggest a OctaveSpecial toolbox, where we put
useful functions that there is a consensus aren't provided by TMW and are widely
used (strjoin, for example, and some smarter i/o functions).  There should be
approval by the core dev's if something gets added to this toolbox, though.

This isn't helpful, but maybe its interesting.  Once I get octave tip to compile
on Freebsd, I will do whatever I can to help ...