"Nick Sabalausky" <seewebsitetocontac...@semitwist.com> wrote in message news:jnimdk$1t1t$1...@digitalmars.com... >> >> It was never a serious project unfortunately, just an experiment. >> > > After having done the experiment, do you feel it's a promising approach? > Are there things you figure would be better done differently? >
I actually found it a lot easier than expected. Most of the constructs that survive to the glue layer are c constructs, which naturally map very well to c code. Writing the actual code is quite straightforward, it's very similar to what is done in 'toCBuffer' for di generation. I never hit a roadblock, so it should be easy for someone else to continue on from there (as Adam has). That said, reconfiguring gdc or ldc to output c (or to target the intended platform) is probably a better idea than writing a new c backend. Either way, druntime needs to be ported.