01:33 dalek joined 02:07 lizmat joined 04:43 jdv79 joined
jdv79 how can it happen that UTF8_ACCEPT but codepoint to grapheme is not ready? 04:46
06:33 domidumont joined
jdv79 nevermind. i partially figured out the issues. 06:39
06:50 domidumont joined, brrt joined 06:54 domidumont joined 07:15 brrt joined 07:18 zakharyas joined 07:32 domidumont joined 07:54 lizmat joined 08:41 cognominal joined
konobi jnthn: queue.acm.org/detail.cfm?id=2927301 09:08
complete with perl tool! 09:09
=0)
jnthn konobi: We already include something a bit like that in the profiler output. :) Except I mind them easier to read when inverted, so it's an icycle graph rather than a flame graph. :) 10:57
timotimo icycle graphs are cool
also, iCycle is obviously an apple product 10:58
nine Certainly cooler than flame graphs ;)
konobi jnthn: it supports icycle 11:00
jnthn haha...the apple bike :P 11:01
konobi: Cool ;-)
Guess I also like cold things much better than hot things, so... :) 11:02
psch ponders how apples hardware lock-down would look on a bike
psch .oO( "damn, another flat tire, guess i can throw this one away too..." )
timotimo can't get a flat tire, because the whole bike is one piece of aluminium 11:03
tires, seat, light... :P 11:04
nine And it's super flat, including the seat
timotimo :D :D :D
konobi jnthn: there's a bunch of support features, etc. mentioned towards the end of the article
nine Might hurt a bit, but it sure looks cool
timotimo what do you call bikes in english that you lie on instead of sitting?
konobi jnthn: it's also important to profile the code in place to see if the code is making libc calls inappropriately (for example) 11:05
timotimo because, imagine one of those, except it's as flat as a macbook
psch and if you can't reach the pedals it's because you're sitting on it wrong /o\
timotimo :D 11:06
psch nevermind that those are also flat on the rest of the flat bike vOv 11:08
nwc10 timotimo: recumbent 11:13
timotimo ah, yes
11:54 andreoss joined 12:07 andreoss left 12:48 colomon joined
dalek arVM: bbf44ea | timotimo++ | src/jit/graph.c:
add jitting for read_fhs
 13:35
timotimo i'm considering grabbing the line-based coverage thing again and building a thing that'll cope with how the core setting is concanetaned 13:36
-- just pushed the rebase 13:45
13:46 domidumont joined
timotimo oh snap, there is already a line_based_coverage_2 13:46
dalek arVM/line_based_coverage_3: b7093ea | timotimo++ | / (11 files):
first prototype of a per-line coverage reporter
 13:47
arVM/line_based_coverage_3: 9923a4b | timotimo++ | / (6 files):
add missing files, adjust op to not use spesh slot

since instrumented frames are actually not fully speshed, and can't easily be, we have to rely on a literal int64 as a pointer to the line report store and allocate it without involving the GC, so it won't move.
arVM/line_based_coverage_3: 2edfd20 | timotimo++ | src/instrument/line_coverage.c:
properly skip over jumplists
timotimo ladida
dalek arVM/line_based_coverage_3: review: github.com/MoarVM/MoarVM/commit/2e369d1851
4408b2a | timotimo++ | tools/parse_coverage_report.p6:
arVM/line_based_coverage_3: a54f8fa | timotimo++ | src/main.c:
show MVM_COVERAGE_LOG in moar --help
arVM/line_based_coverage_3: 9e86701 | timotimo++ | tools/parse_coverage_report.p6:
grey out lines that don't have annotations in the bytecode
nine Somehow the notion of manually numbered branches in a version control system seems just wrong 13:49
timotimo well, it's because i don't want to force-push over my rebases 13:50
ilmari why not? if the old branch is obsolete by the rebase 13:52
timotimo *shrug*
it's a thing i've just started to do at least a year ago. i could force-push instead
14:10 colomon joined 15:00 FROGGS joined 16:35 lizmat joined 17:46 domidumont joined 18:51 brrt joined 19:12 brrt left 21:00 lizmat joined 23:39 lizmat joined

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!