r? @Kobzol

rustbot has assigned @Kobzol.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

I have to say that I'm often clueless when reviewing these kinds of PRs :) Do we have some documentation about what does it mean to update this? Do we even document somewhere which version of emscripten we use for given targets, or is it just an "internal implementation detail".

I have to say that I'm often clueless when reviewing these kinds of PRs :) Do we have some documentation about what does it mean to update this? Do we even document somewhere which version of emscripten we use for given targets, or is it just an "internal implementation detail".

We do not, but we should, probably. I can add the relevant details for our platform support pages?

I think since we build the SDK, it shouldn't have a significant consequence? (aside from the bugfix, of course)

I am not entirely sure what the consequences of this are, tbh. I'm also not entirely sure this won't break Pyodide for @juntyr, either, given 3.1.68 is not 3.1.46, but I'm not sure what to make of emsdk's versioning. It could be SemVer, in which case this should be fine, but then why did 3.1.42 break ABI?

Perhaps @alexcrichton or @Manishearth might know?

Updating the Emscripten version to something newer is definitely a good idea. But it should go hand-in-hand with some documentation (I'll submit a PR later) on this coupling between Emscripten and Rust that won't matter in most cases until you get a weird bug, and how to work around it with -Zbuild-std.

I think I was experiencing this issue for a long while but simply never noticed. Pyodide can use ~2GB RAM before anything breaks, and up until a few days ago even the senseless overallocation during file reading never hit the boundary. So it was by pure chance and additional up-front memory pressure that allocating an extra 1.8GB was no longer fine and I noticed the ABI mismatch.

Thanks for helping to support the Emscripten target @workingjubilee! I really appreciate it.