[ruby-talk:7366] GUIs for Rubies

Thought I'd switch the subject line to the subject at hand.

Some summary observations on recent discussions....

The Mozilla stuff looks like it would involve C++ interface code, which 
tends to be more troublesome to support on many systems, relative to plain 
old C. (This is one of the drawbacks to wxWindows.) 

Mozilla's XPFE is still a work in progress. End-user documentation seems 
scarce. (I haven't done an exhaustive search however.) As yet, there is 
not a track record of successful use by Perl and Python--not that we 
should blindly follow them, but as a smaller community with fewer total 
resources at our disposal, we should think twice about being on the 
bleeding edge.

It seems that many Ruby extension developers apparently don't like dealing 
with C++, meaning thinner support and backup resources if we went with 
things like Mozilla, wxWindows, or gtk--.

GTK+ is pure C, and some substantial fraction of Ruby/GTK+ is *already* 
*done*. And there are already a couple of *books* on GTK+ programming. And 
there is a pretty nice open source GUI-based GUI builder available for it 
(i.e. Glade), unlike the one for wxWindows, which is proprietary.

In addition to Linux, cross-platform support for GTK+ seems tolerably good 
on Windows, and pretty good on major Unix systems. (This condition will 
probably rapidly generalize to the new "UNIX-Mac"--if this hasn't already 
happened.)

None of the GUI options will support as many platforms as Tk, so there 
will be some trade-off of platform range versus GUI capabilities versus 
existing scale of GUI developer community size. What would be best for 
some small percentage of Ruby users on relatively rare (%-wise) platforms 
could considerably disadvantage the great majority of Ruby users on the 
commonest platforms. I think Windows-only or Linux-only or Unix-only 
solutions are each out of the question, but the union of these is probably 
not too far from the sort of quasi-optimal trade-off region that would be 
the best for greatly boosting Ruby's suitability as a heavy hitter, 
GUI-wise (as measured by the maximum potential market for increased Ruby 
usage).

So, for Dave's proposed purpose of closure-seeking, I (provisionally) 
think GTK+ would be the best choice for the time being (relative to all of 
the above considerations). I think that we also want something that (1) is 
already in fairly widespread use and that (2) already has published 
documentation (i.e. books, althought of course we will want to produce 
Ruby/* documentation for whatever choice we make). Because of the previous 
work that has already been done on GTK+, this is one of the least 
resource-intensive and thus fast-to-market GUI solutions as well. 

(I think that we will probably eventually want to support Mozilla a couple 
of years down the road, when it is much more mature, and when we have much 
more resources available to deal with it. However, I think GTK+ will 
remain very viable for many years to come, and will probably remain a 
richer GUI environment.)

Conrad Schneiker
(This note is unofficial and subject to improvement without notice.)