i read and (believe i) understand your points about needing to avoid hacks for a systems language since it will come back to bite you cf. the region tiger's tail. nevertheless, i wonder if there are parts of the overall language that could gain some ground with a 'good enough' approach. dunno, just wondering:
On Tue, May 15, 2012 at 12:23 PM, Jonathan S. Shapiro <[email protected]> wrote: > These are known as unboxed types. Automatic stack allocation is a good > thing, but it should be done generally. What you need to do this is escape > analysis, which is *not* a simple inference problem in the general case. > Conservative escape analysis is often good enough in practice. as an offshoot of looking for something like BitC, thinking about what might be "good enough" for doing e.g. games, i think i would like to have more manual controls over the GC and less magic like escape analysis that i have to rely on -- just because it seems like the former can give the tools to be successful sooner than the latter. sure, i'd like everything to magically work and to let me write code w/out really having to manually think about or manage memory, so if a perfect compiler shows up that's great. i just wonder what hacks-in-the-good-sense we could have in the meanwhile. e.g. random thought, being able to say "please check to see if you can gc any of the things that were heap allocated during this function call" could be nice. _______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
