The ATL parser is generated by ANTLR, and is unfortunately quite slow. Other ANTLR-generated parsers (xText, EMFText, ...) are not doing much better, so there doesn't seem to be an easy way to fix this. The only work-around is to keep the size of your ATL files down. You can do this by splitting up your ATL files into separate modules, and use module superimposition.
I achieved a superimposition of modules, by extracting parts of my "main-module" to several other modules.
The transformation works apart from the content-assist. Is it possible to enable the content-assist for the modules in the superimposition ?
Unfortunately not, as that would require parsing all the other files just for syntax completion information.
Technically, it is possible to do syntax completion based on the new EMFTVM bytecode format (this has been done with the SimpleOCL and SimpleGT experimental languages); EMFTVM bytecode loads much faster than ATL source code. However, doing that would require a rewrite of ATL's syntax completion algorithm, which is currently based on ATL source code.
thanks for your help. I've just an other simple question to
How may I enable 'debuggin' for the superimposed atl-modules? I found nothing
in the documentation about it. If I set my breakpoints to one ast-node of my superimposed atl-module, the debugger never stops there.