Re: [Texmacs-dev] TeXmacs gwt front-end

[Henri Lesourd]

Carl Witty wrote:

> On Fri, 2006-12-01 at 23:49 +0100, Henri Lesourd wrote:
>  
>
> > But it's a long, long way till it becomes as fast as a thin
> > layer over X-Windows. As a matter of fact, if you can't
> > compile your JavaScript and run the corresponding executable
> > file inside your browser, it will *never* happen that it gives
> > us the computing power we need. As for me, I'm not aware
> > of such plans of implementing JavaScript compilation in
> > the main browsers. But it's a classical thing to do, so browser
> > implementors will unavoidably think about that, at some
> > point. But when ?
> >    
>
> The Mozilla project has started on this (based on Adobe's donation of
> the JIT-based virtual machine from Flash Player 9): 
>  http://www.mozilla.org/projects/tamarin/
> The roadmap says "...to be released in 2008."
>
>  
Aha. Interesting.


> Other options for doing intensive computation in a web page include Java
> (or other JVM-based languages) or Flash Player 9 (which can be targeted
> by ActionScript or Haxe).  I think your best bet for the full TeXmacs
> experience in a web browser would be to port TeXmacs to Java, and run it
> as an applet; this is a huge amount of work!
>
>  
This is just not realistic (although not completely infeasible). The
information you gave me, as well as the previous thread discussing
with Amir leads me to think that the Ajax-based implementation of
an X-Windows-like layer would definitely be the way to go, as far
as porting TeXmacs (and other X-Windows/MS-Windows interactive
apps) to the web is cincerned. Something that would provide :

1/ The ability to detect & reprogram events ;
2/ The ability to draw & to display fonts ;
3/ The ability to create subwindows ;

Currently, such a project appears to be feasible, but it is
hard to obtain a very efficient/stable result. But given what
you, along with others say, it seems definitely possible that
at some point, the efficiency of such an "X-Web-Windows"
implementation becomes comparable to what can be done
in C++ on top of a classical Window manager. It is even
rather probable, finally.


> On the other hand, I'm actually in the middle of writing a
> TeXmacs-inspired browser-based editor; I may have an alpha version in a
> few months.  This will display the document using DOM manipulation.  It
> will be missing many of the nicer features of TeXmacs (no CAS
> integration, no fine typography, no blue outlines showing which blocks
> the cursor is in), but I think it may still be usable.
>  
I'm aware of people who tried similar implementations
of such editors inside a browser. All these projects started
and worked well, but they all stopped at some point because
of the lack of efficiency / stability in the DOM, and because
of the lack of portability, too (although this latter point was
never the main blocking point : but already, developing
e.g. on top of Mozilla appears a big challenge in comparison
to what you can do with C++ and native bitmap screen access).

My opinion remains that it is very good for people to reinvent
the wheel by themselves : it often provides them with an
invaluable learning experience. Therefore, I strongly encourage
you in your project.

This being said, I would also say that the kind of "X-Web-Windows"
project I extremely briefly sketched above would be of the higher
interest, not only for porting TeXmacs to the web, but also for
porting lots of other interactive-windowed apps to the web. It
is rather likely that if it could be developed in a sufficiently significant
way, such a project would really raise a lot of interest (much more
than TeXmacs itself probably, due to the general-purpose nature
of such a project).


Best, Henri