TMI: Infinite loops take a long time

[tim]

Hmm, today was exceptionally lacking in productivity! Except for one thing that will hopefully have a very positive long-term effect on productivity, but that's nothing to do with Rust.

Made a few edits to the POPL paper. I should talk to Niko when he's here about what the priorities are.

Finally finished the minor library change (adding some uses of classes to core::comm) that I've been working on for eons, which meant finishing the accompanying destructor fixes and getting a snapshot together. I don't even. Still waiting for the bots to finish building the snapshot. I don't want to say it's done until the snapshot is officially done and I can run tests locally with the new snapshot.

Decided not to check in my fix for issue 2242 -- I had an easy fix to resolve that makes performance much worse, and it seemed like there's no point in coming up with a better fix when Patrick is almost done rewriting resolve and that might fix the bug from scratch anyway.

Fixed issue 2642 -- totally straightforward except for when I couldn't figure out why the test case that I added for it runs forever, and it was because... it contains an infinite loop, and I didn't add any code to bypass it when it actually runs. ^_^ (So many things would be easier if we just compiled code and didn't execute it.)

That's more or less it. Have resorted to typing ps -ef repeatedly while logged into the only buildbot that hasn't finished making the snapshot, to see whether it's running a command that's late in the build process.