japhb Is there a way from the Rakudo level to know when a GC pass happens, or to know how many nursery and full GCs have occured since the last check/program start? It seems like the kind of thing Telemetry would support, but I don't see it. 01:12
01:45 leont left
Geth rakudo: 0ecd037b7d | (Nick Logan)++ (committed using GitHub Web editor) | src/core/Kernel.pm6
[js] Use uname op for sysname and release

Uses the new uname op for sysname and release info. version and machine are not exposed in node, and thus still shell out to uname.
02:06
02:26 dalek left 02:49 Geth left 08:21 dalek joined, ChanServ sets mode: +v dalek, synopsebot left, p6lert_ left, p6lert joined, synopsebot_ joined, Geth joined, ChanServ sets mode: +v synopsebot_, ChanServ sets mode: +v Geth
lizmat japhb: Telemetry could support it, if it was somehow available through nqp:: ops 09:17
timotimo someone want to bikeshed an API for this? :) 09:19
lizmat I would say, similar to nqp::getrusage, but with a type int as the first parameter 09:21
nqp::stats(type,int @array)
if the type is out of range, then simply don't touch the array
this would allow "later" versions of Telemetry to run on earlier versions of the VM 09:22
and also, it would allow other backends to gradually support it (in the beginning it could just be a stub) 09:23
Files=1254, Tests=87994, 393 wallclock secs (21.13 usr 6.60 sys + 2819.55 cusr 234.68 csys = 3081.96 CPU) 09:31
timotimo what would be passed as type for example? 09:33
lizmat an int 09:36
(constant) 09:37
timotimo ah, so there would be nqp::const:: for, for example, STATS_GC? 09:41
i can imagine stats with different "output" types, such as "how many of each type were deleted in the last gc run" (which incidentally i'm just now implementing for the profiler) 09:42
though of course having "int @array" as the second parameter isn't a thing we enforce at the nqp compiler level 09:43
and since we only need backward compat, we can just "know" which value needs what type to be passed 09:44
lizmat there's something to be said for forward compat as well, especially wrt to different backends and dual-living Telemetry 09:47
timotimo in that case, you can just pass an untyped array for whenever you don't know what type would be right, and you'd get the right data 09:48
lizmat that'd work for me. but wouldn't that be rather more work on the VM side?? Plus add unnecessary overhead on something that you want to be as lean as possible 09:49
to avoid affecting the measurements taken ?
timotimo hmm
lizmat goes afk for a few hours
timotimo mumble mumble fast path/slow path, mumble mumble branch predictor, mumble mumble
not sure, actually
but i'd hope the "put stuff into array to give to perl6 code" is either dwarfed by the work done to collect stats, or getting the stats out is so simple that it's at almost no cost in total 09:50
i'm not sure how something like "how many of each type got deleted during the last collection" would be represented in a simple telemetry output log 10:04
since it's not only one dimension more but the size in the "other" direction isn't known in advance 10:06
11:43 llfourn left 12:03 llfourn joined 12:07 llfourn left 12:26 llfourn joined 12:31 llfourn left 12:58 llfourn joined 13:03 llfourn left
Geth rakudo: 3e51bd4e17 | usev6++ | t/spectest.data
[JVM] Don't run tests for utf16
13:14
roast: 445062c150 | usev6++ | 9 files
[JVM] Skip a couple of dying tests
13:15
13:31 llfourn joined 13:36 llfourn left 13:38 llfourn joined 13:45 llfourn left 13:51 llfourn joined 13:55 llfourn left 14:10 leont joined
Geth roast: f4f8290cec | usev6++ | 8 files
[JVM] Fudge a couple of failing tests
14:19
bartolin hopes the next spectest run will be much cleaner 14:22
14:22 llfourn joined
lizmat bartolin++ # keeping the JVM backend alive 14:23
14:27 llfourn left 14:58 lucasb joined 15:16 llfourn joined 15:21 llfourn left 15:31 llfourn joined 15:32 TimToady left 15:37 llfourn left 15:40 llfourn joined 15:45 llfourn left 15:48 llfourn joined 15:52 llfourn left 16:46 llfourn joined 16:51 llfourn left
[Tux] Rakudo version 2019.03.1-140-g3e51bd4e1 - MoarVM version 2019.03-61-g645a70c0d
csv-ip5xs0.749 - 0.757
csv-ip5xs-206.424 - 6.475
csv-parser22.532 - 22.801
csv-test-xs-200.441 - 0.446
test7.485 - 7.638
test-t1.818 - 1.838
test-t --race0.937 - 1.016
test-t-2029.812 - 29.851
test-t-20 --race10.278 - 11.392
16:56
timotimo hey [Tux]
[Tux] o/
timotimo is something wrong with the "[ast 10 days" graph? it seems to only show one day actually
tux.nl/Talks/CSV6/speed4.html
[Tux] I've been away - no internet connection whatsoever. Did not even *touch* a keyboard 16:58
timotimo ah :)
16:58 llfourn joined
[Tux] tux.nl/Files/guirlande2.jpg 16:58
timotimo cool helmet 17:00
17:03 llfourn left
MasterDuke [Tux]: nice 17:37
AlexDaniel very nice pic :) 18:28
dang it someone please implement a deflap arg in bisectable :) 18:30
c: HEAD gist.github.com/AlexDaniel/861cce5...5d3316474f 18:32
committable6 AlexDaniel, Using file “run.p6” as a main file, other files are placed in “sandbox/”
AlexDaniel, ¦HEAD(3e51bd4): « «exit code = 42»»
AlexDaniel c: 6c gist.github.com/AlexDaniel/861cce5...5d3316474f
committable6 AlexDaniel, Using file “run.p6” as a main file, other files are placed in “sandbox/” 18:33
AlexDaniel, ¦6c (37 commits): « «exit code = 42»»
AlexDaniel of course…
c: HEAD gist.github.com/AlexDaniel/861cce5...5d3316474f
committable6 AlexDaniel, Using file “run.p6” as a main file, other files are placed in “sandbox/” 18:34
AlexDaniel, ¦HEAD(3e51bd4): «RUN 1␤RUN 2␤RUN 1␤BUG␤ in block at sandbox/r2805 line 12␤ in sub anastates at sandbox/r2805 line 7␤ in block <unit> at sandbox/r2805 line 3␤␤RUN 2␤ «exit code = 42»»
AlexDaniel c: 6c gist.github.com/AlexDaniel/861cce5...5d3316474f
committable6 AlexDaniel, Using file “run.p6” as a main file, other files are placed in “sandbox/”
AlexDaniel, gist.github.com/4e4c7bc46e1c37e0a8...4a2b9a8837 18:36
AlexDaniel yea that's clearly a new issue
bisect: old=2018.05 gist.github.com/AlexDaniel/861cce5...5d3316474f
bisectable6 AlexDaniel, Using file “run.p6” as a main file, other files are placed in “sandbox/” 18:37
AlexDaniel, Bisecting by exit code (old=2018.05 new=3e51bd4). Old exit code: 0
AlexDaniel or at least it can only be triggered now by this code…
whatever, let's see what we get
bisectable6 AlexDaniel, bisect log: gist.github.com/4ecfc253cdf3753ca8...fc0f7e53cd
AlexDaniel, There are 3 candidates for the first “new” revision. See the log for more details
AlexDaniel Geth: ver github.com/rakudo/rakudo/commit/7b...619e4ef640 18:40
Geth AlexDaniel, version bump brought in these changes: github.com/perl6/nqp/compare/2019....9-gda96d39
AlexDaniel Geth: ver github.com/perl6/nqp/commit/da96d3...4b5760f902
Geth AlexDaniel, version bump brought in these changes: github.com/MoarVM/MoarVM/compare/2...1-g6c7810c
18:43 llfourn joined 18:48 llfourn left 20:04 robertle left 20:44 llfourn joined 20:49 llfourn left 20:56 entonian joined
Geth rakudo/master: 6 commits pushed by (Paweł Murias)++ 20:57
21:02 entonian left 22:46 llfourn joined 22:51 llfourn left 22:56 llfourn joined 23:28 lucasb left