Re: Multi-pass lcc

>
> 1. What is the recommended way of doing mult-p***** with lcc.
> 2. What is the recommended method to estimate the address space size
> covered by jump instructions (branches or function calls)
>
> Thank you very much.

I don't think there is a "recommended" way to create p***** that will
be incor****ated in a modular way in LCC. My opinion is that LCC often
is a good start in compiler projects for different reasons:
1. you can write a backend for a RISC-like processor relatively easily
2. the intermediate representation (in ASTs and data-flow trees) is
well do***ented.

In order to advance its state into a modern compiler there are several
things.
1. Keep the DAGs (at least they can be used for local scheduling --
let's say list-based)
2. Add emitters: at least one for a linearized version of intermediate
"symbolic" code
3. Add target-independent optimizations (that are not there).

But, first you have to record the lcc IR to a more convenient form,
SSA is what is de facto standard now.

Nikolaos Kavvadias