> Hi Nicolas, > > Ok, I see... I guess then, should I opt to use NekoML to create a > compiler, my first task would be to create something that could be > executed by the NekoVM... On succeding with this, I could invariably > modify my output bytecode language and build my own VM.
It depends what is your goal. If you want to design a language that run on the Neko VM, then using the NekoML language to write the compiler is the best choice. I would even say that *in general* NekoML is the best to write compilers. After, if you want your own VM, you might get rid of Neko generation and directly generate your own bytecode. However I'm not sure why you would need that :) > I've started my journey so far by learning as much as possible about > Backus-Naur Form and parser trees, as well as complete mini grammers and > the like, though the actual construction of such grammers from scratch > is still a little hazy at this point. Still, I now feel ready to tackle > the actual parser / scanner construction. AntLR or Cocol/r had seemed > the logical step, but again, I can't help feeling I may miss something > along the way should I follow these routes. At least if I use Neko I > can follow the compiler construction from beginning to end, then when > the VM / execution cycle project begins, I'll at least have more skills > to bring to the table. It depends what you want to acheive. If your goal is to understand how a VM work and be able to write on then studying Neko is maybe enough, you don't have to write your own. Just trying to modify things here and there should be enough for you to get a good understanding of it. Nicolas -- Neko : One VM to run them all (http://nekovm.org)
