dalek arVM: 45a071e | jnthn++ | src/strings/decode_stream.c:
Fix typo in comment.

The keys are like, right next to each other.
11:55
arVM: 5f76d03 | jnthn++ | src/strings/ (10 files):
Make stream decodes return if stopper was reached.
12:39
arVM: 7d59b00 | jnthn++ | src/strings/decode_stream.c:
Don't do hopeless separator searches.

If the decoder doesn't find the stopper, there's no point in going on a separator hunt with the newly decoded chars. Removes a bunch of the cost associated with reading really long lines.
dalek arVM: b9afd77 | jnthn++ | src/strings/decode_stream. (2 files):
Another big speed-up for reading long lines.

Just skip over buffers that can't possibly contain the separator, so we save on expensively scanning them. With this and the previous commit, the heap profiler reaches its prompt within a second or so when reading in a 25MB heap profile.
15:04
jnthn OK, that makes moar-ha feel nicer :) 15:05
timotimo 25 megabytes? that's *nothing*! :P
moritz moar-ha? moar highly available? :-) 15:06
(I guess "heap analyzer")
jnthn heap analyzer :) 15:07
timotimo: :P
dalek arVM: 4afd7b6 | jnthn++ | src/strings/utf8_c8.c:
Fix utf8-c8 decode crash caused by off-by-1.
17:24
timotimo huh! i thought i had tried that and it wasn't enough 17:27
though, what do i know :)
timotimo jnthn: do you have a good suggestion for how to deal with the safepoint thing? 23:57
i mean the leak that exists because we're doing "free at safepoint" over and over but never actually reach a safe point 23:58