Re: State of AWT

[Brian Jones]

Tom Tromey <address@hidden> writes:

> I've done a lot of AWT merging and some debugging recently.
> 
> It turns out that the Gtk+ peers aren't really complete.  For
> instance, it isn't possible to make an AWT checkbox due to a missing
> native method.  Dialogs don't work.  There are Gtk errors in the file
> dialog and one other place I forget right now.
> 
> The AWT code itself is in so-so shape.  Some stuff works, some stuff
> doesn't.  GridBagLayout is completely missing (nice project for
> someone who wants a moderately complex task), as is most of the
> post-1.1 code.  We basically never use the tree lock, which I think is
> a big problem since it will require a more-or-less complete code
> audit.  (Assuming that the tree lock isn't one of those things that
> Sun thought would be a good idea and then later decided not to use --
> the docs aren't very clear.)
> 
> It's not as bad as it sounds though.  A lot of the hard work is done.
> I've been doing my debugging by compiling the Gtk+ peers with gcj into
> a shared library, and then running the TestAWT program with those
> peers and the JNI native code.  Windows, events, buttons, and the
> layout engines work well enough to let me see the bugs in the other
> code.  This setup is noticeably faster than using the JDK (1.2.2); I
> attribute that to the JDK's use of Motif.

I tried over the weekend to run that program with ORP but without
success.  When I get some more time I'll look into it more.

Brian
-- 
Brian Jones <address@hidden>