Sam Clegg

> This seems like a smart change for the reasons @takikawa mentioned. I personally don't really like using `std::visit` and `overloaded` very much since it is a little complex, and...

I'll mark as help wanted in case anyone with access to arm32 wants to take a stab a fixing this.

> Thanks for the PR @shravanrn ! > > Ideally this would be split into smaller pieces as @sbc100 said, but given that there has been almost no other work...

@keithw .. this is just a WIP .. but perhaps it makes a good stepping stone to #1814.

Can you rebase now that #1897 and #1896 have landed? I hope it should make this change much more precise.

Ok! Marking as auto-merge!

Looks like maybe the new site is at https://www.mingw-w64.org/. How would you feel about opening a PR to update the link?

Thanks for the detailed analysis. I appears that `wasm-decompile` expects that all Var's will have names .. but ApplyNames doesn't seem to guarantee this. I think I agree that if...

Actually it looks like a slight modification to src/generate-names.cc does the trick: ``` bool NameGenerator::HasName(const std::string& str) { return !str.empty() && str != "$"; } ``` if this behaviour isn't...

What is the goal of this change? Is it an optimization or something about being able to share this object between stores? If its an optimization then I would ask...