Re: FLTK

[Klaus Ripke <paul-lua@...>]

On Tue, Jul 05, 2005 at 10:34:21AM -0700, whisper@oz.net wrote:
> Lat to suggest it maybe, but I think the Tk toolkit (from Tcl/Tk) might 
> make a better choice. It's more actively developed and there are some 
> very interesting things coming along for Tk 8.5, like skins (themes) for 
> better platform adaptation.
the thread's (initial) issue was a _light_ toolkit,
probably translating to a few 100K binary size with all
but the base system (like libX11 etc.) statically linked.
Just like FLTK's 70-or-so K "Hello world".

I fell in love with Tk before Lua was even born,
and one can learn a lot about toolkit design from it
(including a few how-not-to lessons), but
a) nowadays it's quite fat, and becoming ever fatter
	(what you call "actively developed",
	meaning addition of naughty gimmicks like skins,
	which are fun, but definitely do not belong to a toolkit;
	this is also the reason why wxwidgets sucks ever more).
	It'd be more sensible to use XUL in the first place.
b) using it with Tcl (like PerlTk and others do) is
	ridiculous, and stripping it from the Tcl is feasible
	(studied that approach quite in depth) but lots of work,
	like really a lot. It's not about some tolua.

I'd love to see somebody go the way to do the real LuaTk,
but still, it would not be the light thing.
Starting from "scratch" (read: a more bare bones basis),
borrowing some ideas and maybe code snippets,
should be easier.


Moreover, times have changed.
I just can't believe people are still implementing
"Tree-Widgets" in C, or even worse C++.
Then they see they are ugly.
Then they are inventing configuration systems.
Just look at the complete mess they did with Gtk 2.
Doing all the layout in Lua based on a well done low
level lib, not much more than SDL, should be much faster,
and certainly much more flexible. Ask the gamers.


salud