splitting off bindings [was: Re: patch to TODO]

[Havoc Pennington]

On Thu, 2006-02-16 at 03:50 +0200, Robert McQueen wrote:
> [snip stuff that sounds good]

All sounds great to me, if no objections then please feel free to go for
it. The only thing I'd be sad about is if we lost some of the test
coverage due to untangling the glib bindings. I think I'd rather leave
those in the tree if it's too hard to disentangle them without breaking
various test stuff and tools.

Hrm, another problem is docs; right now the glib bindings are in the
same Doxygen module as libdbus I think (?), and the tutorial skeleton
covers all the bindings at once. I _would_ like to avoid a "lowlevel
only" tutorial as I think tutorial-type materials should strongly
encourage use of a binding. The lowlevel API is, well, lowlevel, and if
we encourage people to start there they are going to end up whining
about its lowlevelness.

So it'd be cool if binding maintainers would maintain a section in the
tutorial for their binding, or at least put a link in the tutorial to
the canonical web location of their binding docs. (e.g. maybe the
tutorial chapter on Qt just links to the right place on trolltech.com)

If using CVS, we do need to move the bindings by copying files on the
server, to keep history; (note, _copy_ the files then "cvs rm" them in
the main tree). But your suggestion of switching to svn is probably the
better one.

I think there's no need to move the broken/not-recently-maintained
bindings, they could just be cvs rm'd and if someone comes along and
decides to pick them up, they could go back in the history and get the
old stuff.

> P.S. As an aside, if we were wanting to consider something a little
> better than CVS, now would be a good time to do it while we make these
> new modules. I've been using SVN recently and am happy with it as a "CVS
> done right", or we could use any of the new crop of distributed systems.

I'd be in favor of SVN if someone is willing to do the work to migrate
over while keeping history. I've also used subversion a lot recently and
it's certainly a noticeable step up and would be a lot easier than
messing with copying server-side files for splitting up bindings.

I would rather not do the distributed system stuff. I'm not a believer
really, especially for something like dbus. I can believe that they make
sense for a ginormous and fork-happy project like the kernel or an
entire distribution, but for dbus I doubt there's enough advantage to
risk picking the wrong one (it's not really clear which of these systems
will be around for the long haul at the moment...)

> I've been absolutely loving working with darcs but have been running in
> to annoying corner cases with spinning & never terminating a little too
> often to recommend its use here.

You're saying it's hard to understand the behavior of a Haskell
program??? shocking! But at least you know it was proven mathematically
correct, as long as you are willing to wait a couple million years for
it to terminate ;-)

Havoc