Re: [udig-devel] holongate vs. SWT_AWT bridge

[Jody Garnett <jgarnett@xxxxxxxxxxxxxxx>]

Andrea Aime wrote:

> Another advantage of using SWT_AWT is that this way the map component
> could be developed as a swing component with a higher reuse potential in other
> applications (from the gt2 point of view). Yet, having to use 1.5 on linux 
> just for this is a pain... 
Agreed (which I why this discussion is so useful).

Given the tradeoffs I would much rather have bits that could reused at 
the gt2 level, it is kind of looking like we are stuck with a JRE 1.5 if 
we want any kind of performance on linux. (Note this is still compiled 
as a JDK 1.4 project).

One we have a render up we should be able to bench mark the following 
three situtations:
- nested for loops (aka holongate w/ out native blit)
- holongate native code blit
- SWT_AWT bridge

The Java 2D, JAI rendering pipeline remains unchanged (and reuseable for 
gt2 apps), it is just the final step to the screen that is giving us 
trouble as a Rich Client Platform application.

We are also taking a certaint amout of pain using JAI. The installation 
of which would already steer us towards bundling a JRE as part of a one 
click installer. In this respect JRE 5.0 does not offer significant 
great annoyance above and beyond JAI.

Jody