Alon Zakai

Interesting that fix helps, I have no idea why... but maybe it was just close to some limit? Btw, was this with binaryen.js or binaryen.wasm?

> And if the full compatibility is not possible, is the partial compatibility even needed? I think that if we can get embind working nicely here, then the long-term benefits...

About CI, try pushing commits with `[ci skip]` [in the title](https://docs.github.com/en/actions/managing-workflow-runs-and-deployments/managing-workflow-runs/skipping-workflow-runs). > I think that declaring instruction factories can be simplified. Which factories do you mean specifically? (wasm-building, the C...

> I mean the wrappers/bindings (binaryen::Module::block, binaryen::Module::if_, etc.) I see, good question. I'd hope there's a way to automate that with embind but I'm not sure. @brendandahl , any ideas?

(Existing tests cover most of this, but I noticed that unreachable code was missing, so that is added.)

> It would be helpful to split this up into a preliminary PR that just moves existing code around, followed by a separate PR adding the laziness. That would split...

> Can we reimplement the eager version in terms of the lazy version? That would also reduce the amount of new code. That would have been good, yeah, but it...

> I don't think we necessarily need to commit new comments just to delete them afterward, but yes, I think moving things to class scope in an existing PR would...

> What's the difference look like? In eager mode we process blocks as a whole, handling each get and set as we see it, and also bundling gets together when...

Last commit removes the laziness-related things here as requested, leaving only a refactoring that splits up `flow()` and calls it on a `LocalGraphFlower` instance. I do still think that is...