Re: [Hecl-devel] Performance

[David W. <dav...@gm...>]

> Variable passing between hecl and Java seems to be a slow point as expected. I have one command that constructs action events (more or less like Swing ActionListeners and events). That command takes 6 parameters - setting the text, action code, action listener, and various other properties - and pretty much all params are coming in as ObjectThings and being extracted at runtime.

You would win major extra bonus points if you could write up some of
this in a bit of code that's publically releasable so that the rest of
us can test it:-)  If not, I'll see what I can do...

> That method is taking around 300ms the first time. I'm not sure if it's the sheer number of arguments, or the object wrapping.
>
> Interestingly enough, the next several times I run through that same method, it took the execution down to a more manageable 15-20 ms. I'm still digging into the hecl internals, but I'm wondering if this is something about a compile/parse step that only hits the first time since the performance characteristics seem to be consistent.

Hecl does some basic caching of variables and commands, so that's
likely what you're seeing.

-- 
David N. Welton

http://www.welton.it/davidw/

http://www.dedasys.com/