Re: [Anygui-devel] Questions about anygui

[Magnus L. H. <ma...@he...>]

James Thompson <ja...@aj...>:
>
> I was reading the archives and came across the Dec post where Magnus
> said the project is pretty much dead and not to waste time on it.

Unfortunately, that does seem to be the state of things, yes.

> Use something like wx instead.  In our project
> (www.gnuenterprise.org) our forms runtime software allows the user
> to select the front end of their choice.  wx, gtk, qt, win32 native,
> curses, etc, etc.

Right. Very Anygui-like :)

> We already have a common library for our applications but several of
> our apps only work in wx (our designer, navigator, etc).  Now I'm
> tossing about the idea of refactoring our tools to work more along
> the lines of python version of netbeans or eclipse (which has a thin
> GUI wrapper that seems in the spirit of anygui).  So a base GUI
> framework using plugins would replace the separate apps, it would
> also provide me with a base application for in house custom apps.

Souns like a good idea to me.

> In our project lots of people use the native windows driver, the wx
> driver, and the gtk driver.  I'd personnally like to get our qt
> driver up and 
> 
> So a few questions..  Does anygui from cvs.....

Just a note about the CVS code base: It's not really in any kind of
stable state. Several features don't really work as they should for
certain back-ends, there are probably lots of bugs etc.

It might be better to take the code (and ideas) you need from it, and
write a custom library for your application. (Or, if you'd like to
base yourself on Anygui, any patches would certainly be appreciated.)

I just wanted to be clear on that Anygui as it is is not really usable
for any stable or commercial system, no matter what features you
need...

>   allow for the interecption of events on a per keystroke level?

No. But the event dispatching system is quite nifty, and adding this
sort of thing wouldn't be too hard. (The problem is, of course, that
you'd have to add it to all back-ends.)

>   what about events caused by things like opening a dropdown box?

No.

>   support some type of drag and drop?

No.

>   support cut n paste on the various platforms?

Well... Sort of. Anygui is cut 'n' paste-agnostic. It works, but
Anygui doesn't know about it.

>   have more complete API docs?

No, sorry. (You could try to run some automated API doc generator on
parts of it...) If you have questions, I'd be happy to help figure
things out (and perhaps document a few things in the process).

>   allow access to the underlying widget set so that if I wanted to use 
>     the base framework in house w/ custom QT widget I'd be able to
>     write my plugin that uses qt hook that widget in.

Yeah, that should be possible.

>   is it lighter weight than wx?

Hm. I guess, in some ways. It's not really big. On the other hand,
you'd have to have something like Qt or wx in the background, of
course. And Anygui is much slower for many things (such as resizing),
if that matters.

> And why did development stall?  Were there issues that arose that
> made anygui seem like a deadend?  Or was it something else?

Many reasons, I guess. I think the basic, simple reason is enthusiasm
was used up before a tangible result (such as a functioning 0.2
release) could boost it again. The reason for this, I guess, might
partially be the "second system" syndrome -- a full refactoring and
lots of fancy features. Maybe. Also, the amount of development needed
(largely because of all the back-ends and all the manual testing
needed of all of them) was, perhaps, too much compared to the (at that
time) current user base/demand (not all that high).

So... As it turned out, I think Anygui was a successful
proof-of-concept that simply aimed a bit too high.

> What I'm thinking about doing is creating the base application with
> anygui.  

Well... In that case you should be prepared for lots of bugs and
missing features (as I mentioned). It may be that you could get some
developers here help you with specific bugs, but I can't guarantee
anything.

> Then create a plugin that link in our existing widget drivers (as
> they don't allow any input without checking with the back end
> first).  If anygui gives me the access I'd need to the events I need
> then I could eventual phase out the existing widgets w/ anygui
> replacements.

Well, it seems interesting :)

> Take Care,
> James

-- 
Magnus Lie Hetland                   Fall seven times, stand up eight
http://hetland.org                                 [Japanese proverb]