Should index of rethrow index blocks (like branch) or just try blocks?

[lukewagner]

For nested rethrow my initial expectation was that rethrow's label would only index try blocks (effectively "filtering" the block stack for try blocks) rather than simply indexing the block stack. This seems marginally more convenient for producers and it doesn't seem like any trouble for anyone. I don't have a strong opinion here, though.

[KarlSchimpf]

The main argument for using a 'label' is that it is a concept already in WebAssembly. However, the expectation is that the label only makes sense if it corresponds to the try block.

[eholk]

@lukewagner, is the idea that rethrow would bubble up to the next try block? For example,

(try // 2
  (block // 1
     (try // 0
        (catch
            (rethrow 1)))))

would be equivalent to (rethrow 2)?

[lukewagner]

I was imagining that the stack being indexed would only contain try blocks, so in your example the only valid indices to rethrow would be 0 and 1. Really, the difference seems pretty minor either way; I was just asking in case anyone else was thinking the same thing or had a more substantial reason.

[lukewagner]

Thinking about it more, since try blocks are also normal control flow blocks (for the purposes of branching), it seems like any producer would already need to be tracking them in the control flow stack and so using a normal block label does seem appropriate.