RE: gtk2-perl

[Murray Cumming Comneon com]

> From: muppet [mailto:scott asofyet org] 
> what do you consider "complete" versus "almost complete"?  i consider 
> 2.0 to be "complete",

Of course it's very vague. It's just a way to show which bindings are more
actively worked on and used and sensibly maintained. It's mostly a
self-declared thing, though I tend to subject people to some Q and A when
they ask to be called "complete". Hmm, it's bollocks really.

> but we leave out quite a few functions 
> intentionally, and others are left out simply because we either can't 
> figure out what they do or if they are actually needed for app 
> developers.

For particular functions, don't be afraid to ask on a mailing list (probably
not this one). For gtkmm we try to wrap everything, even if we have no idea
how it should be used, because we don't expect to know how everything should
work. Our scripts tell us when we have not wrapped something so we can feel
confident about being a "complete" wrapper.

I have _no_ knowledge of your particular binding, but here are some general
comments:

Well, a large part of being "complete" is freezing your API (and ABI if that
is relevant to your language), I think, so you need to persuade me that you
have done that and that you understand what it means. I don't think an API
should freeze until lots of people have had a chance to use it and tell you
that it needs to change - so I'd expect to see lots of releases before
someone suddenly said their binding was "complete" or "frozen".

I think an odd and even numbered versioning scheme is normal these days. So,
as a bottom line, it's important to note that I wouldn't take any binding
seriously if it broke API (or ABI) in a .0 or .2 or .4 release series. (I'd
personally frown on API addition too, but that's less clear cut.) One of the
"complete" bindings has broken API in .0 release series, but I didn't want
to argue too much.

I'd probably also look for an active mailing list also before marking
something as "complete", because that would show that these issues have
probably been discussed.

>  is there a criterion by which we should be going?  is 
> there a currently-maintained gtk.defs file that could've saved us two 
> months of development?

Both the pygtk and gtkmm .defs files are maintained. However, nobody has
done much work for GTK+ 2.4 yet, I think.

> also --- what does it take to become an "official" binding?  
> certainly 
> perl should be on that list if python is. ;-)

See above. If you're "complete" and sane and being used then you're worth
calling official. If there are 2 "complete" bindings for one language then
the "official" one for me will be the most active or most established one.

Murray Cumming
murrayc usa net
www.murrayc.com