[Laszlo-dev] lazlo4java ideas
gbevin at uwyn.com
Wed Sep 21 14:20:10 PDT 2005
this is a great initiative!
> would be slower for rendering and I think that, indeed, it could
> even be faster.
I'm sure it will and the memory handling and garbage collection would
be much better.
> Another big advantage: with a java implementation of laszlo at
> hand, one could easily develop builder tools so it'd be easier to
> defeat Flex et al.
> How do we perform efficient animations in java using Java2D? Do
> animation frameworks already exist on this
http://www.jgoodies.com/freeware/animation/index.html is worth a look
> platform? Should we use them, or implement a strategy from scratch?
... but I only skimmed the docs, never really used it though.
> Does rhino need the laszlo objects to implement a specific
> interface. If so, is byte-code instrumentation a good idea (e.g.
> javassist) ?
I have quite some experience with ASM for bytecode manipulation, so
if you need help there, shoot!
> What portions of the existing compiler can be reused? Validation
> for instance is probably a good bet. Besides, is javac the only
> choice for compiling plain java generated code. Or are there other
> strategies (e.g. javassist, again). Or maybe everything could be
> evaluated at runtime? Or? or?
I think that the trip though javac should be avoided, it's very slow.
I used to do this for a template engine that generated java source
code, which it compiled to bytecode. I converted this to direct
butecode generation with ASM and it was really magnitudes faster. It
wasn't that much more difficult to do.
> How to be XP and still be rigorous with respect to design
> principles? A project like this requires that every aspect be taken
> into account. But at the same time we all know that development
> works only by small increments...
> What would be the most natural milestones that would be compatible
> with the "grand vision"?
> Is it a good idea to target an intermediate level of implementation
> where views would be only static (i.e. no animation) so they could
> serve as building blocks for would-be builder tool projects?
I think it's best to initially start from hello world style examples
and getting them to work from the original LZX source code. The
component object model should first be implemented I think, since
most other things rely on it. After that, it's a matter of what
people feel comfortable with and what they're exited about to implement.
Geert Bevin Uwyn bvba
"Use what you need" Avenue de Scailmont 34
http://www.uwyn.com 7170 Manage, Belgium
gbevin[remove] at uwyn dot com Tel +32 64 84 80 03
PGP Fingerprint : 4E21 6399 CD9E A384 6619 719A C8F4 D40D 309F D6A9
Public PGP key : available at servers pgp.mit.edu, wwwkeys.pgp.net
More information about the Laszlo-dev