Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_log/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by Zoffix on 27 July 2018.
MasterDuke huh. i rebased nqp master onto my default_to_int branch just so it wouldn't get too out of date, but now it won't build, dying on the very first file with `Bytecode validation error at offset 108, instruction 18: operand type 64 does not match register type 32 at gen/moar/stage2/nqpmo.nqp:853 (src/vm/moar/stage0/nqpmo.moarvm:compose)` 00:12
Geth rakudo: xelak6++ created pull request #2188:
Prevent REPL from writing to filehandle when opening a file.		 00:27
MasterDuke oh, it isn't only a rebase which is a problem. i reverted the rebase and it still won't build. i guess it's the new moarvm that's the real culprit
and i get the same thing trying to make new bootstrap files 00:28
anyone have any suggestions? 00:29
timotimo: ping 00:43
timotimo pong 00:48
MasterDuke i feel you might have an idea re my recent question
timotimo hum, a "make clean" doesn't fix it? 00:50
MasterDuke nope 00:51
timotimo o_O
i don't seem to remember changes that required a rebootstrap 00:53
i mean, perhaps an op has changed one of its register types? 00:54
MasterDuke my rename of getstrfromnames
timotimo can you look up what 32 and 64 are, respectively?
renames shouldn't make a difference, as long as the op code number is the same
MasterDuke how do i find the op code number?
timotimo i'd just go into gdb and print MVM_OP_getstrfromnames 00:55
but it's more interesting what 32 and 64 are as register kinds
00:57 Kaiepi left
MasterDuke No symbol "MVM_OP_getstrfromnames" in current context. 00:58
same with "MVM_OP_strfromnames"
how do i see the register kind? 00:59
timotimo looking for defines for that number i guess 01:01
MVM_operand_ and friends perhaps? 01:02
MasterDuke looks like 32 is MVM_operand_int64 01:05
and 64 is MVM_operand_obj 01:06
timotimo ok, maybe by breakpointing we can figure out what the opcode is that it's trying and failing to verify 01:09
MasterDuke val=0x7fffffffd600 in validate_operand at src/core/validation.c:364 01:10
is there some other info i need? 01:12
timotimo i think so 01:16
could it be found in src_cur_op? 01:17
or cur_op maybe?
aha, cur_info should have all info, like ->name
so val->cur_info->name perhaps?
MasterDuke (gdb) p val->cur_info->name $1 = 0x7ffff77164f6 "sp_guard" 01:19
timotimo that is *not* supposed to be in a precompiled file 01:21
i mean, the validator only runs on code that's loaded from a file
and those aren't allowed to have spesh ops in them
what's the opcode for sp_guard? perhaps it's confused by offsets and is seeing an operand as an opcode instead 01:22
like, off-by-one or two bytes
MasterDuke (gdb) p MVM_OP_sp_guard $2 = 807 01:23
timotimo hum 01:26
m: say 807.base(16)
camelia 327
timotimo m: say :16("27")
camelia 39
timotimo i'm not seeing what's going on 01:27
01:28 astj joined 01:29 p6bannerbot sets mode: +v astj
MasterDuke i do get slightly different values for offset and instruction in the error message after i rebase master 01:32
01:37 bray90820_ joined, p6bannerbot sets mode: +v bray90820_ 01:38 bray90820_ left 02:30 AlexDaniel joined, p6bannerbot sets mode: +v AlexDaniel 02:31 AlexDaniel left, AlexDaniel joined, barjavel.freenode.net sets mode: +v AlexDaniel, p6bannerbot sets mode: +v AlexDaniel 02:45 bairdmich joined 02:46 p6bannerbot sets mode: +v bairdmich 02:47 bairdmich left 03:09 Kaiepi joined, p6bannerbot sets mode: +v Kaiepi 03:14 sud012 joined, p6bannerbot sets mode: +v sud012 03:18 sud012 left 03:19 FROGGS left, FROGGS joined 03:20 FROGGS left, FROGGS joined 03:21 p6bannerbot sets mode: +v FROGGS 03:28 FROGGS left 03:29 FROGGS joined, p6bannerbot sets mode: +v FROGGS 03:32 FROGGS left, FROGGS joined 03:33 p6bannerbot sets mode: +v FROGGS 03:42 FROGGS left, FROGGS joined 03:43 p6bannerbot sets mode: +v FROGGS 03:45 FROGGS left 03:46 FROGGS joined, p6bannerbot sets mode: +v FROGGS 03:56 KleinerMann joined, KleinerMann left, mfa29818 joined 03:57 p6bannerbot sets mode: +v mfa29818, mfa29818 left 04:15 sst6621 joined, sst6621 left 05:05 FROGGS left 05:06 FROGGS joined, p6bannerbot sets mode: +v FROGGS 05:51 FROGGS left 05:52 FROGGS joined, p6bannerbot sets mode: +v FROGGS 05:54 FROGGS left, FROGGS joined 05:55 p6bannerbot sets mode: +v FROGGS 06:07 robertle joined 06:08 p6bannerbot sets mode: +v robertle 06:18 FROGGS left, FROGGS joined 06:19 p6bannerbot sets mode: +v FROGGS 06:33 Kaiepi left 06:34 PufferBot joined 06:35 p6bannerbot sets mode: +v PufferBot 06:41 FROGGS left, FROGGS joined 06:42 p6bannerbot sets mode: +v FROGGS 06:44 Kaiepi joined 06:45 p6bannerbot sets mode: +v Kaiepi 06:59 Pici22 joined 07:00 p6bannerbot sets mode: +v Pici22
Geth nqp: 031a78dffe | é™ˆę¢“ē«‹++ (committed using GitHub Web editor) | tools/build/gen-version.pl
satisfy warnings
 07:01
07:06 Pici22 left 07:07 travis-ci joined, travis-ci left 07:09 Vlad14 joined, Vlad14 left
[Tux] Rakudo version 2018.06-333-gf8b79306c - MoarVM version 2018.06-380-gac3d3c769
csv-ip5xs1.046 - 1.081
csv-ip5xs-207.619 - 7.882
csv-parser25.231 - 27.237
csv-test-xs-200.453 - 0.474
test10.177 - 10.311
test-t2.343 - 2.354
test-t --race0.983 - 1.007
test-t-2041.583 - 42.224
test-t-20 --race13.470 - 13.594
 07:24
07:33 ascheel14 joined, ascheel14 left 08:05 Ven` joined 08:06 p6bannerbot sets mode: +v Ven` 08:22 baweaver16 joined 08:23 p6bannerbot sets mode: +v baweaver16 08:24 baweaver16 left
Geth star: 9feb960ef2 | (Steve Mynott)++ | tools/star/release-guide.pod
doc manual steps needed in rakudo.org repo
 08:28
08:55 crayon2 joined 08:56 p6bannerbot sets mode: +v crayon2, crayon2 left
lizmat Files=1247, Tests=75961, 377 wallclock secs (16.20 usr 5.68 sys + 2680.93 cusr 239.19 csys = 2942.00 CPU) 09:14
09:19 Ven` left 09:21 Ven` joined 09:22 p6bannerbot sets mode: +v Ven` 09:40 robertle left 09:41 robertle joined 09:42 p6bannerbot sets mode: +v robertle 09:45 Zord10 joined 09:46 p6bannerbot sets mode: +v Zord10 09:51 Zord10 left
Geth rakudo: 07d00db148 | (Alexius Korzinek)++ | 2 files
Prevent REPL from writing to filehandle when opening a file.

This is a fix for issue #2184.		 10:46
rakudo: 7c08ba7443 | (Zoffix Znet)++ (committed using GitHub Web editor) | 2 files
Merge pull request #2188 from xelak6/master

Prevent REPL from writing to filehandle when opening a file.
synopsebot RAKUDO#2184 [open]: github.com/rakudo/rakudo/issues/2184 [easy to resolve][good first issue][regression][āš  blocker āš ] open() fails in REPL
10:50 astj left 10:52 robertle left 10:54 robertle joined 10:55 p6bannerbot sets mode: +v robertle, DrForr left 11:03 DrForr joined, astj joined, p6bannerbot sets mode: +v DrForr 11:04 p6bannerbot sets mode: +v astj 11:08 astj left 11:21 nate joined 11:22 p6bannerbot sets mode: +v nate 11:23 nate left 11:28 Ven` left
tbrowder_ o/ #perl6-dev 11:42
iā€™m still looking for someone to discuss pod with. o
rakudo GH #1893 has an interesting slant. as iā€™ve been looking at moving pod to a new slang, why does it have to be part of the main compiler at all? treat it like whitespace and, as TisonKun says, use another program to handle the pod. 11:46
seems like that would: (1) make p6 parsing faster, (2) use less memory, and (3) make pod easier to treat properly. 11:49
ilmari tbrowder_: aren't there metamethods that give you access to the pod?
tbrowder_ yes, if i understand your question, but they still come from the main p6 compiler. TisonKun and i think a separate program should do the parsing and provide the equivalent of the current $=pod object. 11:52
11:52 robertle left
tbrowder_ the main compiler would ignore pod. 11:52
so it would still have to have some pod knowledge, but not nearly as much as now. 11:55
i donā€™t have a clue if that is feasible, but it sounds to me like it could work. 11:57
or maybe the current method would be better if it would have a separate pod slang but ignore the pod unless an optional p6 switch was turned on. 12:05
12:07 Zoffix joined, p6bannerbot sets mode: +v Zoffix
Zoffix tbrowder_: you mentioned going to the printers regarding Intro. Perl 6 brochure. Note that it's size is A5, not A4, and US equivalent is "Half US Letter" 12:08
tbrowder_ ah, okay, thanks! 12:09
Zoffix And 19 out of 20 pages will have to be edited to make it fit to US format natively. 12:10
12:11 Ven` joined, p6bannerbot sets mode: +v Ven`
tbrowder_ i understand. from the one i printed i think a little scaling on half-letter wonā€™t look too bad. 12:12
maybe print it on colored paperā€”weā€™ll see what magic the printer can do. 12:13
Zoffix
.oO( 80lb gloss )
 12:14
tbrowder_ ok. is the indd file the adobe source? 12:15
Zoffix Yes
tbrowder_ thnx
have you tried scribus on an indd file? 12:18
Zoffix No, and I highly doubt it would work.
Doesn't even come up in the list of supported files; at least with Scribus 1.4.1.svn (which is ancient) 12:20
tbrowder_ apparently if the indesign prog saves the file as .ps or .eps scribus can use it. 12:21
Zoffix What about IDML? 12:23
EPS, it just creates 20 separate EPS files, one for each page
And they aren't even editable. Just imported as an image 12:25
Ah, nm, if you ungrop it ungroups it, but all the text is single letters grouped together
tbrowder_: this article claims newer Scribuses can open IDML: opensource.com/article/18/7/adobe-...urce-tools 12:29
tbrowder_: here's an IDML of the brochure: temp.perl6.party/Introducing-Perl6-...BLEED.idml 12:30
It'll probably ask for links to fonts/images, those are in the marketing repo
Document fonts and Links: github.com/perl6/marketing/tree/ma...1516098660 12:31
Zoffix &
12:31 Zoffix left 12:47 tharkun26 joined 12:48 p6bannerbot sets mode: +v tharkun26, tharkun26 left 12:49 a3f_ joined, p6bannerbot sets mode: +v a3f_ 12:50 a3f left, camelia left 12:52 shareable6 left, nativecallable6 left 12:54 reportable6 left, benchable6 left, statisfiable6 left, greppable6 left
tbrowder_ thank you :-D 12:54
12:56 camelia joined
tbrowder_ btw, if you saved as .ps it should just ceate one file afaik, an .eps is by design to be inserted into .ps or other compatible format filed. 12:57
12:57 p6bannerbot sets mode: +v camelia, robertle joined 12:58 p6bannerbot sets mode: +v robertle 12:59 sulvone8 joined, sulvone8 left 13:11 sekizo3 joined
robertle all the marketing material discussions reminded me of something: where can I get a camelia sticker? 13:11
13:11 p6bannerbot sets mode: +v sekizo3 13:13 sekizo3 left
robertle if none exist already then I'll volunteer and get some printed... 13:14
timotimo liz and woolfy have stacks of 'em 13:15
they bring them to every conference they attend
robertle awesome
timotimo there are different stickers as well as buttons in multiple sizes
the one i like most has kind of an extra tab on the outside of one of the wings that says "perl6" or "perl6.org"
lizmat robertle: if you would come to the Amsterdam.PM meeting later today at TechInc 13:33
we could get you some and then some other Perl 6 paraphernalia :-)
lizmat now goes outside to spend ~ 2 hours in a hot car on a hot road 13:34
13:34 lizmat left, Ven` left 13:35 Ven` joined 13:36 p6bannerbot sets mode: +v Ven` 13:50 reportable6 joined 13:51 p6bannerbot sets mode: +v reportable6 14:00 skids joined 14:01 p6bannerbot sets mode: +v skids 14:14 pmurias joined 14:15 p6bannerbot sets mode: +v pmurias
pmurias jnthn: what are the benefits of passing around a global context on the jvm backend instead of storing stuff in static variables? 14:15
14:15 Zoffix joined, p6bannerbot sets mode: +v Zoffix
Zoffix releasable6: status 14:15
releasable6 Zoffix, Next release in ā‰ˆ11 days and ā‰ˆ4 hours. 4 blockers. 81 out of 335 commits logged (āš  9 warnings)
Zoffix, Details: gist.github.com/f00d714671af95d89e...f43bb00b66
Zoffix Reminder that the release is coming up and there are some blockers that need to be resolved.
14:16 Zoffix left
jnthn pmurias: I think that multiple instances within the same JVM were possible 14:26
pmurias: Which is used to run the spectest faster
I don't recall that being passed around though, I thought just the threadcontext was
Geth nqp: 488ba3c2e6 | (Jonathan Worthington)++ | 2 files
Allow specification of repr with `stub` declarator
 14:34
nqp: 96f58597fc | (Jonathan Worthington)++ | src/NQP/Actions.nqp
Whitespace correction
|Tux| Rakudo version 2018.06-335-g7c08ba744 - MoarVM version 2018.06-383-g67a9afef6
csv-ip5xs0.950 - 0.958
csv-ip5xs-207.804 - 8.206
csv-parser24.144 - 24.568
csv-test-xs-200.448 - 0.514
test8.708 - 8.954
test-t2.159 - 2.218
test-t --race0.945 - 0.977
test-t-2038.101 - 38.343
test-t-20 --race12.677 - 13.109
 14:37
looks like running it on the box while actually using it as a desktop causes some slowdown
2018-08-06 13:34:27 test-t 2.205 14:38
2018-07-17 13:19:48 test-t 2.203
2018-08-03 16:57:31 test-t 2.197
2018-08-03 17:01:13 test-t 2.175
2018-08-07 16:28:38 test-t 2.159
jnthn Hm, 2.159 would be a new low, yes? :)
|Tux| that is why the second table is posted: it is the top-5 14:39
14:39 travis-ci joined, travis-ci left
jnthn aha :) 14:39
pmurias jnthn: that's passed on the threadcontext
jnthn wonders if he can get it below 2 :) 14:40
|Tux| knows 14:42
I think it might get a lot under 2 if Ā«break;Ā» could be optimized to a goto (or if there is something like break that is not an exception) 14:43
I think it might get a lot under 2 if Ā«next;Ā» could be optimized to a goto (or if there is something like next that is not an exception) 14:44
(sorry)
my mainloop has about 50 next's
jnthn That already happens in some cases :) 14:46
(Not looked at if the opt is done in your particular case)
|Tux| 34, but still, a lot 14:47
14:48 brrt joined, p6bannerbot sets mode: +v brrt 14:59 FROGGS left 15:00 FROGGS joined, p6bannerbot sets mode: +v FROGGS 15:01 FROGGS left, FROGGS joined 15:02 p6bannerbot sets mode: +v FROGGS
timotimo i think it requires the loop body to be inlined into the containing frame that has the loop's control structure? 15:03
for that, we may want to slightly alter the size conditions for inlining; if the containing frame is pretty small, but the contained one is very big, we should say "ok, that's fine. it's as if we were inlining the outer frame into the inner one" 15:04
15:09 ABIX_Adamj joined 15:10 p6bannerbot sets mode: +v ABIX_Adamj 15:11 ABIX_Adamj left 15:15 diakopter left 15:28 brrt left 15:29 davidfg413 joined, davidfg413 left, ollien12 joined 15:30 p6bannerbot sets mode: +v ollien12, ollien12 left 15:36 diakopter joined, p6bannerbot sets mode: +v diakopter 15:58 lizmat joined 15:59 brrt joined, p6bannerbot sets mode: +v lizmat, p6bannerbot sets mode: +v brrt 16:00 Geth left 16:01 Geth joined, ChanServ sets mode: +v Geth, p6bannerbot sets mode: +v Geth
Geth nqp/speshplugin_guardstaticcode: 2f1cf4fbd6 | (Timo Paulssen)++ | src/vm/moar/QAST/QASTOperationsMAST.nqp
bind speshguardgetstaticcode

requires github.com/MoarVM/MoarVM/pull/932		 16:10
16:11 robertle left
Geth nqp: timo++ created pull request #495:
bind speshguardgetstaticcode		 16:12
rakudo/speshplugin_guardstaticcode: 5b881e9f24 | (Timo Paulssen)++ | src/vm/moar/spesh-plugins.nqp
teach "assign" spesh plugin about Proxy

requires github.com/perl6/nqp/pull/495		 16:13
rakudo: timo++ created pull request #2189:
teach "assign" spesh plugin about Proxy
rakudo/tune-hash: 656b763d8d | (Jonathan Worthington)++ | src/Perl6/Metamodel/BOOTSTRAP.nqp
Generalize attribute default mechanism

So that we'll be able to use it for cases that are not Scalar, e.g. making the storage of Hash automatically be an nqp::hash instead of having to do repeated checks of that.		 16:26
rakudo/tune-hash: 0b1e6d54f7 | (Jonathan Worthington)++ | 2 files
Make Map/Hash $!storage always init to nqp::hash
rakudo/tune-hash: 99f7f4cc85 | (Jonathan Worthington)++ | 2 files
Tune Map/Hash for $!storage changes, and more

This is mostly a case of kicking out nqp::isconcrete instructions, but we can also use some the new and cheaper instruction for making a Scalar container with an initial value.
16:26 FROGGS left 16:27 FROGGS joined, p6bannerbot sets mode: +v FROGGS 16:30 [TuxCM] joined 16:31 p6bannerbot sets mode: +v [TuxCM] 16:44 robertle joined, p6bannerbot sets mode: +v robertle 16:48 Ven` left
Geth rakudo/tune-hash: 1f65bf243f | (Jonathan Worthington)++ | 3 files
Don't have p6scalarfromvalue check for definedness

Since we now always promise we will always provide a descriptor that is defined to it. Results in rather more compact code.		 16:49
rakudo/tune-hash: 37c7e6e558 | (Jonathan Worthington)++ | 2 files
Further redundant check removals in Map/Hash
timotimo nothing is certain but desc and taxes 16:54
Geth rakudo/tune-hash: 90f678f3da | (Jonathan Worthington)++ | 2 files
Avoid more unrequired descriptor checks
 16:56
lizmat jnthn: would that also work for QuantHashes having a Rakudo::Internals::IterationSet ? 17:00
jnthn: in any case, way cool :-) 17:02
jnthn Probably there's more places it can be applied :)
Please forge ahead and do so; I'm just showing the way on the hottest things :-)
lizmat yeah... can't wait to do that :-)
shouldn't I wait for hte merge ? 17:03
jnthn Yeah, it just passed spectest so I'll merge very soon :)
gist.github.com/jnthn/196263d7e888...b2e42b2008 runs in half the time it used to thanks to that branch :)
lizmat so: "has Rakudo::Internals::IterationSet $!elems" 17:04
is all that it is needed to have it auto-vivify the attrinbute ?
jnthn Ah, no, see the stuff I did in BOOTSTRAP
lizmat :-(
jnthn Well, I mean, that'll viv it to the type object as usual :) 17:05
lizmat well,one of the reasons of using an IterationSet is that I wanted to be able to pass them around without them being HLLized 17:06
otherwise it is just an VMHash repr
jnthn Yeah, you can do that :)
lizmat I can do what ?
jnthn Actually I put in a change to NQP earlier today so we can set up IterationBuffer early enough with the correct repr
diakopter I vaguely remember bumping against that circularity issue a few years ago 17:07
lizmat then I would like to set IterationSet also in there :-)
jnthn I was originally going to look at doing the $!storage trick I've done with Map on List
But it turns out it's a bit harder there, so I figured I'd get it working for the easier case first :)
Geth rakudo/master: 7 commits pushed by (Jonathan Worthington)++ 17:08
17:16 FROGGS left 17:17 FROGGS joined, p6bannerbot sets mode: +v FROGGS
lizmat jnthn: something like "my $other := nqp::getattr(map,Map,'$!storage')" would that vivify the $!storage in 'map' ? 17:18
17:18 Kaiepi left
Geth nqp: 7cd46f0b4c | (Zoffix Znet)++ | src/QRegex/Cursor.nqp
Don't check var until we actually gonna use it
 17:18
jnthn Yes 17:21
Well, at the moment :) 17:22
17:22 Kaiepi joined 17:23 p6bannerbot sets mode: +v Kaiepi
lizmat ok so I can take checks out for that then 17:23
reduce: nqp::defined(nqp::getattr(self,Map,'$!storage')) && nqp::existskey(nqp::getattr(self,Map,'$!storage'),key.WHICH) 17:24
to:
nqp::existskey(nqp::getattr(self,Map,'$!storage'),key.WHICH) 17:25
jnthn Darn, getting array assignment below the inline size is tricky
I can sorta kinda get it close-ish
lizmat jnthn: so I could do that type of reduction ? ^^
jnthn lizmat: Should be able to, yes
lizmat ok, will do that then for object hashes 17:26
[TuxCM] Rakudo version 2018.06-342-ga2c34504a - MoarVM version 2018.06-383-g67a9afef6
csv-ip5xs1.018 - 1.154
csv-ip5xs-207.598 - 7.851
csv-parser24.676 - 24.937
csv-test-xs-200.467 - 0.491
test8.720 - 8.801
test-t2.178 - 2.269
test-t --race0.938 - 0.949
test-t-2038.391 - 39.521
test-t-20 --race12.465 - 13.006
 17:29
(on request)
Kaiepi PufferBot, help 17:33
PufferBot Kaiepi, address me with 'build', 'test', 'spectest', or 'stresstest' to test building Rakudo, running tests, and running Roast's test suite respectively on openbsd. Address me with 'all' to attempt to run a full build with all tests.
Kaiepi PufferBot, all
PufferBot [openbsd] Running complete Rakudo build and tests (this will take a while)...
jnthn Ah, goodness, the return type check handling is still rather bulky in the code it spits out
Not to mention that it partly duplicates the kind of checks that decontrv does
Geth nqp: 5b87dc34b1 | (Zoffix Znet)++ | src/HLL/Grammar.nqp
Micro-opt method LANG

  - Get rid of variables we don't really need
  - Most of LANG calls don't use extra args, so swap the always-slip
   version to a ternary that makes argless calls ~2.2x faster,
   while slowing down argful calls within noise (~1.5%)		 17:34
Kaiepi would it be worthwhile to include nqp tests for RakudoBot?
PufferBot [openbsd] Failed to build Rakudo and run all tests... See the output at fpaste.scsys.co.uk/581365 17:35
jnthn notices the time and goes home for dinner :)
bbl
AlexDaniel Kaiepi: heh that looks bad. Is everything alright with the bot setup? 17:37
17:38 travis-ci joined, travis-ci left
Kaiepi looks like i might need to remove the quotes in the flags i use 17:39
17:41 PufferBot_ joined
Geth nqp: f8d218659b | (Zoffix Znet)++ | tools/build/MOAR_REVISION
[MoarVM Bump] Brings 3 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...3-g67a9afe 67a9afe Have sp_fastcreate do a direct nursery allocation 2f36e26 Add branch hint macros to nursery allocation 7507090 Make spesh thread more GC-responsive		 17:42
Kaiepi PufferBot_, all
Ā¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...3-g67a9afe
rakudo: a6edf511d8 | (Zoffix Znet)++ | tools/build/NQP_REVISION
[NQP Bump] Brings 6 commits

NQP bump brought: github.com/perl6/nqp/compare/2018....3-gf8d2186 f8d2186 [MoarVM Bump] Brings 3 commits 5b87dc3 Micro-opt method LANG 7cd46f0 Don't check var until we actually gonna use it 96f5859 Whitespace correction ... (7 more lines)
17:42 p6bannerbot sets mode: +v PufferBot_
rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....3-gf8d2186
6e661e9e4f | (Elizabeth Mattijsen)++ | src/core/Hash.pm6

Because of work on the tune-hash branch
17:43 PufferBot_ left, PufferBot left 17:44 PufferBot joined
Kaiepi PufferBot, all 17:44
PufferBot, help
hm
17:45 p6bannerbot sets mode: +v PufferBot
Kaiepi ah that's it 17:45
it's building atm
PufferBot [openbsd] Failed to build Rakudo and run all tests... See the output at fpaste.scsys.co.uk/581366
17:46 travis-ci joined, travis-ci left, PufferBot left
17:47 See joined, See left 17:50 [TuxCM] left 17:51 FROGGS left 17:52 FROGGS joined 17:53 p6bannerbot sets mode: +v FROGGS
Kaiepi i'll check if the build fails when i try it manually 17:56
17:56 Eduard_Munteanu2 joined 17:57 p6bannerbot sets mode: +v Eduard_Munteanu2 17:58 Eduard_Munteanu2 left 18:11 FROGGS left, FROGGS joined 18:12 p6bannerbot sets mode: +v FROGGS 18:14 [TuxCM] joined, p6bannerbot sets mode: +v [TuxCM] 18:18 pheizax joined, p6bannerbot sets mode: +v pheizax, pheizax left 18:19 PufferBot joined, Alex`26 joined, p6bannerbot sets mode: +v PufferBot, p6bannerbot sets mode: +v Alex`26
Kaiepi PufferBot, all 18:20
PufferBot [openbsd] Running complete Rakudo build and tests (this will take a while)...
[openbsd] Failed to build Rakudo and run all tests... See the output at fpaste.scsys.co.uk/581367 18:21
18:21 BeastieBot joined 18:22 p6bannerbot sets mode: +v BeastieBot
Geth rakudo: 0f1b36a7a5 | (Elizabeth Mattijsen)++ | src/core/Stash.pm6
Greatly simplify Stash.package_at_key
 18:22
Kaiepi BeastieBot, all 18:24
BeastieBot [freebsd] Running complete Rakudo build and tests (this will take a while)...
18:24 Alex`26 left
Kaiepi ok so running ./Configure.pl on openbsd manually works, so it's an issue with the bot 18:25
odd though, it didn't used to do that
now that it's built manually though... 18:27
PufferBot, all
PufferBot [openbsd] Running complete Rakudo build and tests (this will take a while)...
Kaiepi if i can afford it, i might set up a netbsd vps for another bot like this as well 18:28
can't use the netbsd vm on my machine unless you're willing to wait a full day for it to finish
Geth nqp: 82790f9892 | (Zoffix Znet)++ | 4 files
Skip trailing whitespace
 18:33
nqp: 5872eda888 | (Zoffix Znet)++ | 13 files
Swap post-inc/dec into pre-int/dec

Timing difference drowns in noise, but by dumping MAST I see there's one fewer `set` ops in pre version. I also changed
  `$x := $x + 1` into `++$x`; dumping MAST for that one shows
the two are identical		 18:34
18:39 travis-ci joined, travis-ci left 18:40 FROGGS left 18:41 FROGGS joined
Geth rakudo: b308db24d3 | (Zoffix Znet)++ | tools/build/NQP_REVISION
[NQP Bump] Brings 2 commits

NQP bump brought: github.com/perl6/nqp/compare/2018....5-g5872eda 5872eda Swap post-inc/dec into pre-int/dec 82790f9 Skip trailing whitespace		 18:41
rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....5-g5872eda
bc2ca535f1 | (Elizabeth Mattijsen)++ | src/core/control.pm6
18:41 p6bannerbot sets mode: +v FROGGS 18:43 Zoffix joined, p6bannerbot sets mode: +v Zoffix, ChanServ sets mode: +o Zoffix 18:44 p6bannerbot left, p6bannerbot joined, Zoffix sets mode: +o p6bannerbot, Zoffix left 18:45 brrt left
PufferBot [openbsd] Failed to build Rakudo and run all tests... See the output at fpaste.scsys.co.uk/581368 18:46
18:48 FROGGS left 18:49 FROGGS joined, p6bannerbot sets mode: +v FROGGS
18:53 FROGGS left, FROGGS joined
Kaiepi last two tests that openbsd failed are easily fixable, first might be a flapper i'd have to check 18:54
18:54 p6bannerbot sets mode: +v FROGGS 18:56 FROGGS left 18:57 FROGGS joined 18:58 p6bannerbot sets mode: +v FROGGS
Kaiepi sorry for asking again, but how do i check which tests are flappers again? 19:00
19:00 FROGGS left 19:01 FROGGS joined
BeastieBot [freebsd] Failed to build Rakudo and run all tests... See the output at fpaste.scsys.co.uk/581369 19:01
19:01 p6bannerbot sets mode: +v FROGGS
timotimo i usually ask on the channel %) 19:02
but if you have the irclogs, you can search "ZOFFLOP"
19:10 FROGGS left 19:11 FROGGS joined, p6bannerbot sets mode: +v FROGGS 19:17 FROGGS left, FROGGS joined 19:18 p6bannerbot sets mode: +v FROGGS 19:21 Harzilein27 joined 19:22 FROGGS left, p6bannerbot sets mode: +v Harzilein27 19:23 FROGGS joined, Harzilein27 left, FROGGS left 19:41 AlexDaniel left 19:42 AlexDaniel joined, AlexDaniel left, AlexDaniel joined, p6bannerbot sets mode: +v AlexDaniel 19:51 x[LGWs4x4i]uG2N0 left 19:55 x[LGWs4x4i]uG2N0 joined, p6bannerbot sets mode: +v x[LGWs4x4i]uG2N0
Geth rakudo: 36d37cbc8b | (Elizabeth Mattijsen)++ | src/core/Hash.pm6
Simplify my Int %h.ASSIGN-KEY
 19:56
19:57 skids left 20:01 benchable6 joined, [TuxCM] left, p6bannerbot sets mode: +v benchable6 20:11 PufferBot left
Geth rakudo: 81a5e4ba89 | (Elizabeth Mattijsen)++ | src/core/Hash.pm6
Simplify my Int %h{Any}.AT-KEY|ASSIGN-KEY|BIND-KEY
 20:15
rakudo: 1cd654a224 | (Elizabeth Mattijsen)++ | src/core/Hash.pm6
Manually inline Hash!AT-POS-CONTAINER

It became small enough to do so now with recent changes.		 20:31
timotimo did you mean AT-KEY-CONTAINER there? 20:32
lizmat argh 20:33
yeah
comment added to commit
20:42 travis-ci joined, p6bannerbot sets mode: +v travis-ci
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Simplify my Int %h{Any}.AT-KEY|ASSIGN-KEY|BIND-KEY' 20:42
travis-ci.org/rakudo/rakudo/builds/413291662 github.com/rakudo/rakudo/compare/3...a5e4ba89fb
20:42 travis-ci left
buggable [travis build above] ā˜  All failures are due to: failed make test (1 failure). Across all jobs, only t/04-nativecall/12-sizeof.t test file failed. 20:42
diakopter lizmat: sorry if it's a dumb question, but doesn't it still need the AT-KEY-CONTAINER method for other stuff 20:45
MasterDuke m: my @a; my Int $b = 2**65; @a[$b] = 3; # huh, i just assumed this would work 20:49
camelia Cannot unbox 66 bit wide bigint into native integer
in block <unit> at <tmp> line 1
jnthn lol, that'd need a lot of memory :D 20:50
timotimo i don't think we can give malloc or mmap more than 64bits for the size ;)
and that array would have 8 times that size, because it stores pointers
jnthn diakopter: It's a private method, so if it were wanted somewhere else, it'd be a compile error 20:51
diakopter m: my @a[9999999999999999] 20:54
camelia Illegal dimension in shape: 9999999999999999. All dimensions must be integers bigger than 0
in block <unit> at <tmp> line 1
diakopter m: my @a[] = BEGIN { @a = () } 20:56
camelia 5===SORRY!5=== Error while compiling <tmp>
An exception occurred while evaluating a BEGIN
at <tmp>:1
Exception details:
Cannot resolve caller STORE(Array: List); none of these signatures match:
(Array:D: Iterable:D \iterable, *%_ā€¦
diakopter m: my @a[] = BEGIN { @a = [] }
camelia 5===SORRY!5=== Error while compiling <tmp>
An exception occurred while evaluating a BEGIN
at <tmp>:1
Exception details:
Cannot resolve caller STORE(Array: Array); none of these signatures match:
(Array:D: Iterable:D \iterable, *%ā€¦
20:59 [TuxCM] joined 21:00 p6bannerbot sets mode: +v [TuxCM]
MasterDuke how much memory would it take? at work i have access to serves with 768gb of ram, and while i don't use perl 6 for stuff, i frequently just stick all the data i'm working on in memory because why not 21:01
lizmat need to go away now, will look at stuff tomorrow& 21:03
21:03 lizmat left
MasterDuke though usually i'm using hashes, i assume they have the same limit to the number of keys? 21:03
jnthn: btw, when you've got some free time, maybe you'll have a suggestion about colabti.org/irclogger/irclogger_log...8-08-07#l1 (tldr, i can no longer build my branch of nqp that defaults to int operations, which incidentally, i wouldn't mind people taking a look at to see if what i've done so far looks reasonable) 21:06
github.com/MasterDuke17/nqp/tree/d...ead_of_num 21:08
21:12 robertle left
MasterDuke timotimo: or if you have any other debugging suggestions 21:14
or anybody else
bueller?... 21:15
21:17 ripdog8 joined, ripdog8 left 21:20 trisk18 joined, trisk18 left 21:29 rej4 joined 21:30 p6bannerbot sets mode: +v rej4 21:31 rej4 left 21:44 nativecallable6 joined 21:45 p6bannerbot sets mode: +v nativecallable6 22:16 pmurias left 22:27 Kaiepi left, Kaiepi joined 22:28 p6bannerbot sets mode: +v Kaiepi
Geth roast: xelak6++ created pull request #461:
Add new test for issue #2158.		 22:28
22:29 bisectable6 joined 22:30 p6bannerbot sets mode: +v bisectable6 22:47 Zoffix joined, p6bannerbot sets mode: +v Zoffix
Zoffix MasterDuke: I'm assuming you have tried full build? Like, with make realclean, perl Configure and so on? For both rakudo and nqp 22:49
Geth nqp: 56f163fa64 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
[MoarVM Bump] Brings 2 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...5-g3e679da 3e679da Merge pull request #934 from xelak6/master 448e75b Get the number of bytes to be processed from the current buffer and not from the header.		 22:50
Ā¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...5-g3e679da
rakudo: 28df1bb0df | (Zoffix Znet)++ | tools/build/NQP_REVISION
[NQP Bump] 56f163f [MoarVM Bump] Brings 2 commits

NQP bump brought: github.com/perl6/nqp/compare/2018....6-g56f163f
MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...5-g3e679da 3e679da Merge pull request #934 from xelak6/master 448e75b Get the number of bytes to be processed from the current buffer and not from the header.
Ā¦ rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....6-g56f163f
Zoffix ZOFFLOP: t/spec/S17-supply/syntax.t You planned 83 tests but ran 74. 22:51
ZOFVM: Files=1303, Tests=152979, 171 wallclock secs (25.43 usr 3.35 sys + 3715.08 cusr 169.06 csys = 3912.92 CPU)
R#2158 now segfaults for me :/ 22:52
synopsebot R#2158 [open]: github.com/rakudo/rakudo/issues/2158 [regression][āš  blocker āš ] :enc<utf8-c8> on Proc appears to drop a lot of content
Geth roast: 8496358550 | (Alexius Korzinek)++ | S32-encoding/decoder.t
Add new test for issue #2158.
 22:53
roast: 44f31e533b | (Zoffix Znet)++ (committed using GitHub Web editor) | S32-encoding/decoder.t
Merge pull request #461 from xelak6/master

Add new test for issue #2158.
22:53 travis-ci joined, p6bannerbot sets mode: +v travis-ci
travis-ci NQP build failed. Zoffix Znet '[MoarVM Bump] Brings 2 commits 22:53
travis-ci.org/perl6/nqp/builds/413348787 github.com/perl6/nqp/compare/5872e...f163fa64f3
22:53 travis-ci left
Geth rakudo: 677f23fb35 | (Zoffix Znet)++ | t/spectest.data
Add S32-encoding/decoder.t to list of files

Added on S#461 github.com/perl6/roast/pull/461 Partially covers R#2158 github.com/rakudo/rakudo/issues/2158
  (the encoding issue is covered, but the OP code in the Issue now
segfaults for me, which the test doesn't cover)		 22:55
synopsebot S#461 [closed]: github.com/perl6/roast/pull/461 Add new test for issue #2158.
R#2158 [open]: github.com/rakudo/rakudo/issues/2158 [regression][āš  blocker āš ] :enc<utf8-c8> on Proc appears to drop a lot of content
Zoffix travis died 'cause: "Error while compiling op strfromname (source text: "nqp::strfromname(~$/)"), no registered operation handler" 22:56
22:59 cheshire_cat joined
Zoffix damn SEGV in t/spec/S32-io/open.t 22:59
ZOFFLOP: t/spec/S32-io/open.t SEGV
22:59 p6bannerbot sets mode: +v cheshire_cat
Zoffix ZOFFLOP: t/spec/S32-io/io-path-extension.t exit status 17 22:59
22:59 cheshire_cat left
MasterDuke Zoffix: well, i never got rakudo to build before, but yeah, for nqp did make realclean, Configure, etc 23:01
#6 get_all_in_buffer (tc=tc@entry=0x555555559c80, ds=0x55555818dcf0) at src/strings/decode_stream.c:481 23:03
#7 0x00007ffff7a6d28a in MVM_string_decodestream_get_all (tc=tc@entry=0x555555559c80, ds=<optimized out>) at src/strings/decode_stream.c:495 23:04
from ` perl6-gdb-m 'run("curl", :enc<utf8-c8>, :out, "-s", "www.google.com").out.slurp.chars.say'`
Zoffix Comment on github.com/rakudo/rakudo/issues/21...-411230232 23:05
I get "Program received signal SIGSEGV, Segmentation fault. 23:21
0x00007ffff704e4d8 in _int_free (av=0x7ffff737a620 <main_arena>, p=<optimized out>, have_lock=0) at malloc.c:3996
3996 malloc.c: No such file or directory.
23:21 astj joined 23:22 p6bannerbot sets mode: +v astj
timotimo that smells like earlier memory corruption 23:23
Zoffix How did MasterDuke get better output?
timotimo running the program with valgrind can often give more details
that comes from a gdb session
oh, you are in gdb
you need to "bt" and if you want a lot of spam, "bt all"
no 23:24
MasterDuke Zoffix: is your moarvm built with --debug=3?
timotimo "bt full" is the command
Zoffix MasterDuke: I think so
timotimo to have fewer "<optimized out>" you'll also need --optimize=0
MasterDuke valgrind is throwing errors i've never seen before
timotimo :D
MasterDuke it won't even run the code
timotimo oh? 23:25
MasterDuke oh, it won't even run 'say "hi"'
Zoffix I got --no-optimize
Is that the same?
`--no-optimize --debug` is what I use
MasterDuke gist.github.com/MasterDuke17/563b6...b62adfaac4 23:26
i leave optimization on, but --debug=3
Zoffix same error 23:27
but btfull gives more stuff
MasterDuke let me try clang... 23:28
timotimo bugs.archlinux.org/task/59551
if you want valgrind to work again, downgrade your glibc below 2.28-1
MasterDuke ah, thanks 23:29
Zoffix 4 0x00007ffff7618fcb in MVM_decoder_take_all_chars (tc=0x603a30, decoder=0x7ffff1a06698) at src/6model/reprs/Decoder.c:263
result = 0x0
timotimo The Control-flow Enforcement Technology behavior can be changed for SHSTK/IBT at run-time through the "GLIBC_TUNABLES" environment variable.
perhaps with that you can temporarily disable it and make valgrind run again
Zoffix valgrind outpuit: github.com/rakudo/rakudo/issues/21...-411236460 23:38
23:44 Zoffix left 23:54 VM_ joined, p6bannerbot sets mode: +v VM_, VM_ left

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!