Tango vs Phobos

[bearophile]

Robert Fraser:
> LLVMDC only supports D1 and is soon to come out.

D2 is good, I can see only one or two problems in it, so making LLVMDC support D 2.x too looks important to me.

What are the issues left in LLVMDC? I see the LLVM backend as the most (the only, maybe) important for the future of D (more important than the closed and not community-hackable backend of DMD, and the complex GCC), are such issues important? If they aren't important then it may be better for Walter to change the specs of D 2.x to made D 2.x fit for LLMV. Because D is meant as a language easy to compile too, but if has some features that can't be implemented well on free backends like the ones of GCC and LLMV then such features look like fit to be removed/modified, otherwise no fully complaint alternative implementations of D may appear, and this sounds silly.

Two persons have already answered me regarding the idea of LLVM becoming the reference backend for the D language (that Walter too switches to work on), that most people can improve and tune, but I haven't understood the problems they have explained to me.

Bye,
bearophile