Re: [PATCH] Jump Tables for Switch/Case and Computed Goto

[Paige DePol <lual@...>]

On May 6, 2014, at 12:03 AM, Karel Tuma <kat@lua.cz> wrote:

> Excerpts from Paige DePol's message of 2014-05-06 03:26:49 +0200:
>> Huge ivore tower syntax thing

What? I have no idea what you mean by that? It is not something I said.


> Code does not look horrible though I can't find cosmetics slapped
> on otherwise useful/powerful primitive agreeable.

Again, what?


> I know it is simpler to hack the parser than write transpiler
> for this, but the point is to keep C as minimal as possible,
> ie there is power ratio which cosmetic syntax ruins.

O.o  (And I don't use that emote often at all!)


> Same effect, with less code, can be achieved by making current goto
> labels addresable identifiers. ie identifier yielding index in
> internal label table.

Essentially I did make addressable identifiers, that can be any constant value.


>> No jumping outside of scope
>> No jumping when there is a sub-scope peresent

You can jump when there is a sub-scope. You can jump over it, just not into it.


> Unfortunately this fairly limits utility. For actual computed
> gotos, You need to carry with each label BCPos range _from_
> where it is legal to jump to it (literally CAMEFROM Implementation :).
> Technically, move the scope invariant check from compile time to run time.

Why would we need to do that, we know at compile time if all jumpto statements can reach all jump points without violating local scope.

Thank you for your feedback, though I have to be honest, I did not understand half of it! :(

~pmd~