Re: Questions about Bytecode

> This looks as if you intend to write out everything while parsing.
> Consider using multiple sections and tables in the created file,
> prepare them in memory or temporary files, and write out everything
> when finished. Having everything in memory will allow to run the
> parsed source immediately, without creating an precompiled file. Then
> you'll immediately use the structures as required by the VM, apart
> from what comes into mind while only parsing.


So it sounds like this is probably the best approach. As an example,
what would be a typical layout of a compiled file look like.


I'm guessing something like this:


headers (such as maybe cafebabe [as another example] and maybe version
numbers)
literals
ops


----


I'm also wondering how labels are implemented in the bytecode.


And as a final side note, is it more common to see one stack or two
stack machines. What benefits do they have?

Return to the comp.compilers page.
Search the comp.compilers archives again.