Joe Bryan

@bacwyls you're just running out of memory while trying to copy out to the home road (the crash handler itself is crashing, the leaks are happening there and are not...

Instead of directly targeting `next/kelvin/411`, this can be built on top of #6859 (until it and the relevant vere pr are merged).

just remove the `.vere.lock` file from your pier and restart (the lockfile implementation is very basic at best ...).

These container constructors exist for two reasons: ergonomics in "casual" programming, and various wet-gate shenanigans. In normal code, it's best to just cast lists and use `+gas:in`/`+gas:by` for sets and...

@Fang- can you post the relevant types too?

FWIW, I don't think this is worth adding a new name: `(zing (reap n list))` is readable and not that much more inefficient.

As an experiment, I've rewritten the serf to a) stop allocating events, effects, and ipc messages on the home road, and b) keep the arvo kernel on an inner road...

@pkova this has to be coordinated with the linked jet changes to avoid introducing new jet mismatches.

After trying for a bit, I can't find a way to take advantage of it. Barring some legitimate use case, it's a bug/oversight, imo.

@xiphiness here's the "jub" variant i proposed in #6672: ``` ++ job |* [b=* c=$-((unit) *)] => %= . b `_?>(?=(^ a) p.n.a)`b c `_?>(?=(^ a) |~((unit _q.n.a) q.n.a))`c ==...