Matt Diephouse wrote:
Dan wrote an entry on his blog yesterday entitled "WWIT: Universal bytecode":http://www.sidhe.org/~dan/blog/archives/000421.html In it, he talks (surprise, surprise) about being able to run bytecode across machines without the need of a compiler. If you look at the comments, I asked what this meant for PMCs. As it currently stands, many of the languages currently targeting Parrot would need their PMCs distributed in order to run any programs written in that language.
Well, if you have some mixed environment, you'd probably build parrot on all machines with the PMC's needed. Something like:
perl Configure.pl --with-tcl --with-python to get these PMCs built on it.
Are there any thoughts from the new powers that be? Should PMCs be avoided when possible? Should there be PIR versions of the PMCs available for distributing bytecode? Should as much functionality as possible be put into the core PMCs? I'd like the ability to distribute bytecode without PMCs; I think it's something worth working for. What's the best way to make this work in light of PMCs?
I still like to unify ParrotObjects and PMCs more. An Integer PMC shouldn't really be much different then a instantiated object of a closed ParrotClass with one attribute - the native int value. When that is done, there is no physical difference between a builtin PMC and a class built in a PIR module, so that you can mix these two depending on their availability.
leo
