dalek roast/6.c-errata: a80962e | (Zoffix Znet)++ | S03-operators/mix.t: 00:17
roast/6.c-errata: Restructure file to avoid triggering SPESH bug
roast/6.c-errata:
roast/6.c-errata: This commit makes no changes to tests themselves. It merely
roast/6.c-errata: avoids using one set of variables for the entire test file, and
roast/6.c-errata: insted chops it up into smaller chunks, with (duplicated) sets of
roast/6.c-errata: those variables in each chunk.
roast/6.c-errata:
roast/6.c-errata: This is needed after the merge of the set operator work[^1], as
roast/6.c-errata: it triggers a SPESH bug[^2] in MoarVM that crashes the test run.
roast/6.c-errata:
roast/6.c-errata: [1] github.com/rakudo/rakudo/pull/934
notviki github.com/perl6/roast/commit/a809...9f064de278 00:18
It looks scarier than it is. The code was just indented. 00:19
(I was hoping github would show that; alas)
samcv scarier? 00:31
oh
you mean less line changes that it looks
git's diff has much to be desired
dalek kudo/nom: cb9ec03 | (Douglas Jenkins)++ | src/core/IO/Socket/INET.pm:
IO::Socket::INET. Fixed bug where IPv6 URIs were not parsed correctly when passed as host:port (E.g. "[::1]:3000"). Split new method into a multi method with type constraints. Now fails when invalid family or port values are passed. Added constants for MIN and MAX port values
 00:36
kudo/nom: 64de09a | (Douglas Jenkins)++ | src/core/IO/Socket/INET.pm:
Moved host-port splits to own function. Replaced optional new args with %rest parameter. Changed type constraints on port/family to Int(Cool).
kudo/nom: df20d8b | (Douglas Jenkins)++ | src/core/IO/Socket/INET.pm:
fixed bugs with uri parsing
kudo/nom: fbd061b | (Douglas Jenkins)++ | src/core/IO/Socket/INET.pm:
Removed ":D" from Str type on :$localhost. :$localhost is no longer "!" required. Checks if :$localhost is undefined, and if so doesn`t try to split it.
samcv well i'm off to continue working on full unicode 9.0 support for Moar :) 00:38
WOOO GOT IT WORKING AGAIN 00:50
dalek kudo/nom: 37941d9 | (Zoffix Znet)++ | src/core/IO/Socket/INET.pm:
Use consistent code style
samcv prepending working now :)
notviki FWIW, I checked all modules in ecosystem and all provide a port to IO::Socket::INET, so won't be affected by the removal of iffy default. 01:10
samcv wooo notviki i am down to only 7 failed unicode graphpme tests! 02:01
notviki sweet
samcv does a dance 02:02
for some reason Extend + letters or a space finds 2 chars though. but using numbers works fine. have to see why the hell 02:03
other than that just failing ones that will require us to save state and track more than 2 characters 02:04
maybe some special emoji path idk
but i don't think emoji use Extend characters or if it's just part of the test?
hmm
i will have to rework the whole thing to be able to switch modes between by character and by sequences 02:40
i think for now i'm just going to save some state iterating over the codepoints or something. have different return values for prepend or extend vs otherwise 02:51
that should actually speed things up a lot. since atm if we seen an extending mark we backtrack to the previous character and 'redo it' 03:04
wait actually may not fix that specific thing at the moment but should allow us to more quickly know without checking properties if we need to extend or not 03:05
notviki, are there any failing tests right now? 05:31
i'm testing my Even_Moar_Unicode_9.0 branch right now that makes much bigger changes so we can pass the other 7. and need to see which were'nt caused by me breaking things 05:32
looks like only two failing atm on moar master. will probably implement all of unicode 9.0 then fix the problems it's having on ENV things 05:55
woooooooooo down to only 3 failing tests :) 05:59
(for unicode grapheme breakup)
nine_web m: say "'I'm still alive" 07:49
camelia rakudo-moar 37941d: OUTPUT«'I'm still alive␤»
nine_web darn...looks like "only" ssh is broken
samcv nine_web, you know MoarVM right github.com/samcv/MoarVM/commit/86c...0667d05738 07:52
can you tell me why this made MoarVM randomly crash before starting
Unhandled exception: should eventually be unreachable; is what I get. and can't gdb cause no stack 07:53
nine_web samcv: wow, that doesn't make any sense 07:56
samcv yeah
i made WAY more changes the commit before 07:57
nine_web samcv: have you tried without the fprintfs?
samcv and i've done make clean then make again
yep
i removed all of them in commits after it
gonna run git bisect anyway just because i have no clue wtf 07:59
nine_web samcv: I see no breakage here after compiling that commit? 08:01
samcv really?
nine_web oh wait...now I get one 08:02
samcv :3
nine_web BUT I did manage to get a backtrace: gist.github.com/niner/c223dcf3c559...9528a23089 08:03
samcv git bisect says 86cafe43aa35f7bbea6990b46321a40667d05738 Implement breaking on the 3rd regional indicator seen is the bad commit 08:04
nice @backtrace
weird but i don't think it's bad or maybe it is. probably
oh well. i have backtrace
nine_web My suspicion is that you broke the short circuiting in the if, causing it to run should_break when it didn't before. And this time on data it cannot cope with? 08:05
samcv maybe
maybe it's is_hangul returning weird values? idk i always return proper values
i didn't write is_hangul though 08:06
nine_web Yeah, nqp builds successfully with: gist.github.com/niner/711ce2944037...d51b3f5861 08:09
But of course, I have no idea if this would break your algorithm or not :) 08:10
[Tux] This is Rakudo version 2016.12-180-g37941d95a built on MoarVM version 2016.12-35-g293bda71 08:11
csv-ip5xs 3.136
test 13.218
test-t 5.300
csv-parser 14.124
samcv hm
yeah nqp is fine
nine_web, what happens if you then build rakudo 08:12
nine_web builds and tests fine. Only test failing is one that gets confused by the debug output on STDERR 08:13
samcv rakudo builds fine?
or nqp?
nine_web yes
samcv which one
rakudo?
nine_web rakudo 08:14
samcv which commit are you on?
nine_web 19df3584e68269f719bcc039c9c41d0f9ca4b793
samcv i mean on my branch though 08:15
that's not a commit i have
nine_web Ah, MoarVM 86cafe43aa35f7bbea6990b46321a40667d05738 08:16
samcv ah kk 08:17
so you still got errors on it though? but could build rakudo?
i'm getting crashes trying to build nqp
that backtrace though MVM_codepoint_to_row_index seems like it's breaking because the row doesn't exist 08:24
but not sure why it would do that
gonna try breaking up that commit into smaller pieces then bisecting again 08:25
hmm nine_web it seems to be breaking due to should_break_mode = should_break(tc, n->buffer[pos], n->buffer[next_pos], should_break_mode); 08:28
and then checking should_break_mode instead of the return value of the function
nine samcv: with my patch, MoarVM, nqp and rakudo build just fine even on MoarVM commit 321b8f3bc3d2010593cb8545d0be89e143419821
samcv which project is that commit for 08:29
or patch
nine This one: gist.github.com/niner/711ce2944037...d51b3f5861 08:30
MoarVM commit 321b8f3bc3d2010593cb8545d0be89e143419821 is Even_Moar_Unicode_9.0
samcv woo!!! 08:33
awesome
applied it to the head that passes 100% tests
nine, also similarly this could maybe be causing problems too: github.com/samcv/MoarVM/commit/29f...9b1dd60265 08:38
cause i was getting some spec failurs
though maybe not because i use another function to convert back or something? idfk
so far everything is passing tho!!! 👍 08:47
oop no now getting some failures :( spoke too soon 08:48
doesn't even run t/spec/S05-substitution/subst.t 08:49
===SORRY!=== Normalization: illegal call to get grapheme
not getting any backtraces there either 08:58
gonna try changing kernels or something
brb for reboot 09:03
arnsholt samcv: If you're dealing with guts, --ll-exception is often quite informative 09:08
samcv arnsholt, as an option for what? 09:11
gcc? moar?
if i do perl6-gdb-m --ll-exception it really doesn't tell me where in moarvm the issue is 09:12
nine samcv: break MVM_exception_throw_adhoc before run in gdb 09:14
samcv nice :)
ah crashes in MVM_unicode_normalizer_get_grapheme 09:15
figures. i was worried about that before i remember
ah weird. gcc didn't fail here compiling 09:21
even though it tries to pass one less argument
yeah that would give it negative values and thinsgs :) 09:22
well MVM_get_ready_status returns the 'old' expected result from MVM_unicode_normalizer_process_codepoint_to_grapheme
which only had 0 or positive, and I use negative to act as 'no, and save this state for the next char' 09:23
arnsholt samcv: It's a rakudo option 09:24
samcv i wonder if unicode 10.0 draft has a test yet, though i could just try random ZWJ sequences
arnsholt But yeah, if it's Moar-level guts, what nine++ said 09:25
samcv gonna grep and make sure no other places it didn't typecheck at :\ 09:26
nice! passed some tests that totally crashed it 09:30
lizmat goo *, #perl6-dev! 11:58
*d
m: say "fake news" if True but False; # my hope for this year 11:59
camelia ( no output )
samcv what True fake news but false? 12:21
m: say "fake news" if False but True; 12:23
camelia rakudo-moar 37941d: OUTPUT«fake news␤»
samcv isn't if True but False just misleading news?
lizmat I guess it's all a matter perspective 12:24
:-)
DrForr liz - Outline for FOSDEM sen. 12:25
psch ...in contrast to an energy perspective :)
DrForr psch: It's all interchangeable.
samcv idk
dalek p: e5ef8f5 | (Pawel Murias)++ | src/vm/js/Operations.nqp:
[js] Fix the return type of the nqp::setinputlinesep and nqp::setinputlineseps ops.
 12:26
p: 2d84e4d | (Pawel Murias)++ | src/vm/js/Compiler.nqp:
[js] Tweak the NYI placeholder.
p: 9149926 | (Pawel Murias)++ | src/vm/js/Operations.nqp:
[js] Add support for the nqp::p6return op in rakudo-js.
p: 44aed3c | (Pawel Murias)++ | src/vm/js/ (2 files):
[js] Stub nqp::backtrace.
nqp: 9f40d1c | (Pawel Murias)++ | src/vm/js/nqp-runtime/co (2 files):
nqp: [js] Fix nqp::isrwcont.
notviki samcv: there aren't any failing tests, but you need bleed perl6 for passing bleed roast, because often a bugfix is acompanied by the test for it. 14:10
samcv yeah was a while since i said that. :) thanks tho
stmuk blogs.perl.org/users/steve_mynott/2...uture.html 14:12
dalek ast: ea5c5c0 | dugword++ | S32-io/socket-host-port-split.t:
Added test for resolved bugfix RT #130474 (#211)
 14:13
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130474
dalek kudo/nom: 6ffbbf4 | (Zoffix Znet)++ | t/spectest.data:
Add S32-io/socket-host-port-split.t to list of files to run
 14:17
ast: 38a3cf8 | (Zoffix Znet)++ | S32-io/socket-host-port-split.t:
Add standard header and footer to test file
 14:18
notviki NeuralAnomaly: status 14:37
NeuralAnomaly notviki, [✘] Next release will be in 2 weeks and 5 days. Since last release, there are 57 new still-open tickets (56 unreviewed and 0 blockers) and 180 unreviewed commits. See perl6.fail/release/stats for details
tbrowder can zef be made the primary installer (i.e., can it replace panda in the whole Rakudo Perl 6 system)? 14:41
moritz considers making a Rakudo distribution just like Star, but with zef, and a clearer vision 14:42
the vision part is what's lacking right now 14:43
it might be "only the compiler, and enough to get other modules installed"
notviki we already have a Rakudo distribution just like Star, but with zef..... 14:44
moritz: blogs.perl.org/users/steve_mynott/2...uture.html 14:45
arnsholt I dunno. "Just compiler and zef" might be a good addition to the distribution-space 14:47
Kind of a bare minimum distro, but one step up from just the compiler 14:48
notviki Sounds like a good way to even further increase confusion for users. 14:49
Especially when all of this gets package as distro packages 14:50
samcv ^
+1
there should be some easy thing for distros to do to package us i think
arnsholt I guess. OTOH, I think much of the onus for confusion (or lack thereof) in terms of users must fall on the Linux distro packagers 14:51
Most notably, they need to be at least moderately clueful about how Perl 6 works in terms of distributions 14:52
RabidGravy yeah it strikes me that they will package all the bits separately anyway 14:53
arnsholt Yep. And different distros have different packaging philosophies
dalek kudo/nom: 48e8ccc | samcv++ | src/core/Cool.pm:
uniprop, for Bidi_Mirroring_Glyph return empty string if no BMG

Before we would return a U+0000 character; returning an empty string is will evaluate as Falsey, unlike a U+0000 in a string.
arnsholt I'm not sure (Linux) distros having individual packages for different (Perl 6) distros is a terribly useful way to do things 14:54
samcv you mean for rakudo and rakudo star yada yada? 14:55
arnsholt Yeah
samcv but a blocker for debian was that they had panda, but none of its dependencies 14:56
which was stupid. shouldn't have even put it in the repo then
arnsholt I think it'd make more sense for distros to package the rakudo compiler, and whatever modules (or not) they want in the base package
And then other modules as separate packages
RabidGravy totally 14:58
and if for some reason they want to have rakudo star as a package it's just make it some meta thingy that just depends on rakudo and the modules in star 14:59
ralisch They will probably also make the "rakudo" package recommend the "star" package, so if a noob installs just "rakudo" they still get something that works. 15:01
tbrowder whatever, can't zef at least be part of all the packages released by the p6 crew? including the travis one? 15:02
samcv buggable, release 15:04
buggable, help
buggable samcv, tags | tag SOMETAG | eco | eco Some search term | speed
samcv ahh 15:05
what do i type
notviki FWIW, one improvement I'd like to see in Rakudo Star is predictable scheduling of releases. I don't care if it's monthly, every third month, or once a year. When someone asks when's the next release and you can't even ballpark it is annoying.
stmuk: ^
And considering it's the only user-facing distro we have, essentially what ends up happing is people asking "when's the next release of Perl 6" and you go "I've no idea" 15:06
arnsholt Yeah, that sounds sensible 15:07
samcv what do we usually do? like a week or two after the last normal release?
arnsholt But it'll require ata least some additional work to make sure the bits that are packaged actually work properly together
samcv plus like 'whenever'
notviki samcv: that's the problem, there's no "usual"
samcv k 15:08
notviki Some time it's every 3rd month, sometime it's monthly
samcv: what do you type for what? I think those are the only commands buggable has
samcv and doesn't change for actual real life reasons?
oh. when the next release is
notviki NeuralAnomaly: status 15:09
NeuralAnomaly notviki, [✘] Next release will be in 2 weeks and 5 days. Since last release, there are 56 new still-open tickets (55 unreviewed and 0 blockers) and 181 unreviewed commits. See perl6.fail/release/stats for details
notviki samcv: our compiler release date doesn't change for actual real life reasons. If one person can't release it another does the release.
Same should be for star
samcv i think should be every 2 months for star 15:10
1 is too low. 3 seems too high
stmuk notviki: I'm aiming for .01, .04, .07 & .10 15:19
notviki Noted. Thanks.
buggable: star releases :is: Estimated Rakudo Star releases for 2017: .01, .04, .07 & .10 15:20
huggable: star releases :is: Estimated Rakudo Star releases for 2017: .01, .04, .07 & .10
huggable notviki, Added star releases as Estimated Rakudo Star releases for 2017: .01, .04, .07 & .10
stmuk but if there are BSD or Windows problems with anything large its delayed
samcv are we shooting for 3 months? 15:21
stmuk thats basically what happened last year (except there would have been two releases delayed) 15:22
.10 was broken with Windows and .07 with some BSDs 15:24
samcv those version non-stars you mean? 15:25
stmuk both star and non-star 15:26
ralisch Is roast being run on them? 15:27
notviki Not on every OS 15:28
There's a bunch of failures on Windows actually: github.com/perl6/roast/issues/197 15:29
Though many are just bogus tests
(like ones that don't switch the terminal to utf8)
samcv notviki, RE rt.perl.org/Ticket/Display.html?id...et-history what file is it in?
notviki samcv: github.com/rakudo/rakudo/blob/nom/...#L131-L146 15:30
samcv so the perl 6 pod is supposed to let you put anything you want there?
notviki Well, my point in the ticket was that likely it shouldn't 15:31
And we shouldn't parse config with colonpair
samcv are you only allowed keys and values there? or can it be any code you wish? 15:33
notviki .ask TimToady is Pod supposed to allow arbitrary code to be run? For example, right now I can stuff a BEGIN block into table :config and run anything I want. Is that the idea or should it be just quoted text? RT#130477 15:34
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130477
yoleaux2 notviki: I'll pass your message to TimToady.
samcv also variables could be useful there maybe
not sure about use case atm though 15:35
ralisch If variables are interpolated into POD, the docs would be automatically updated when invariants or constants are changed. 15:39
cognominal Speaking of release, is there a target date for 6.d and feature freeze date in 6.d ? 15:42
notviki cognominal: nope
samcv there are proposed things though cognominal github.com/perl6/specs/blob/master/v6d.pod 15:43
notviki There's lots of work that needs to be done to make 6.d possible.
samcv ^
cognominal I have seen this page. Certainly nothing major in it to justify an incompatible release. I just wanted to know if there was a policy. Probably good to align it with a major feature, possibly unrelated like good JIT to machine code. 15:49
stmuk 6.d is a lang spec and nothing to do with MoarVM 15:50
notviki or Rakudo for that matter 15:51
cognominal: but that list is not the lots of things that need to be done to make 6.d possible. We currently don't have any sane infrastructure to support two language versions and all that needs to be implemented .
Some relevant docs: github.com/rakudo/rakudo/blob/nom/...ersions.md 15:52
stmuk there was a multiple settings branch 15:53
notviki docs site also needs a way to mark stuff as available in language version XYZ 15:54
stmuk and p6doc 15:55
cognominal stmuk : I know but in the past, Larry Wall release a new version of Perl to align with his book. We don´t always suck in marketting 15:56
*released 15:57
notviki is dubious of 6.d happening in 2017
cognominal I just want to be able to give some kind of answer when asked :) 15:58
notviki Tell them it'll be on Diwalli, but don't tell which year ;)
We've done this before :P
cognominal and got a lot of flak until the release 15:59
samcv it'll be done when it's done 16:00
also why do we need 6.d unless we make incompatible changes though
cognominal on the other hand, it is good to see the page for 6.d so skinny. Either it means people did not got around to it or it means there is no hurry.
samcv Split the Moar-fudged test files up into the fudged tests (which go in speculation) and the passing ones (which go into 6c). 16:02
what is this implying not to put it in if it doesn't pass rakudo?
stmuk isn't 6.c complicated enough? :)
samcv yes
ralisch There's no hurry because people had a lot of time to think what they wanted in the language already. Just because 6.c was released doesn't mean people will start thinking of things to break already. 16:03
jnthn Note that language versions aren't just - or even primairly - about backwards-incompatible changes, but also about giving people a way to declare in their code that they need a certain new set of things.
cognominal jnthn++ 16:04
jnthn Including the hundreds of small things that got nailed down by bug fixes.
(Granted those would we handled by a 6.c.1)
cognominal but that would be a minor version
dalek kudo/nom: 2782836 | samcv++ | docs/language_versions.md:
2016 is in the past, update the language_versions.md to reflect this
 16:06
jnthn And yes, 6.d will also be a marketing opportunity, which will be a consideration for its timing also. :) 16:07
samcv 6.c.1 sounds weird though
6.c⅒ 16:08
why not
notviki 6.e would be a good marketing opportunity ^_^
AlexDaniel
.oO( let's slap one extra combining character for each minor version )
ralisch samcv: So, that's what we're gonna do today, bikeshed. Is that it?
samcv heh 16:09
just make it even crazier. 6.c.½.∋.¿
each one must be from a different type
AlexDaniel wanted to provide a demonstration of that, but then noticed notviki…
notviki heh 16:10
I have a terminal that doesn't break but I can't get it to look as nice as terminology
ralisch Which is it?
notviki gnome terminal and xterm 16:11
samcv why not 6.c.a or something. idk
or are we only going with minors
ralisch I'm using solarized in my xterm
samcv the major will never change though so. idk
notviki I don't even know how to change the settings in xterm
ralisch They go into ~/.Xresources 16:12
notviki ain't got that file
cognominal the good thing it that we can do 6.c.aa if there are more than 26 minors :)
samcv eww xterm
bad unicode
notviki m: say 6.c.aa after 6.c.z
camelia rakudo-moar 278283: OUTPUT«No such method 'c' for invocant of type 'Int'␤ in block <unit> at <tmp> line 1␤␤»
notviki m: say v6.c.aa after v6.c.z 16:13
camelia rakudo-moar 278283: OUTPUT«False␤»
notviki cognominal: we can't :)
samcv m: say 'v6.c.aa' cmp 'v6.c.z' 16:14
camelia rakudo-moar 278283: OUTPUT«Less␤»
AlexDaniel
.oO( Perl 7 )
m: say v6.c.aa cmp v6.c.z
camelia rakudo-moar 278283: OUTPUT«Less␤»
samcv m: use nqp; say ORDER(nqp::unicmp_s('v6.c.aa','v6.c.z')) 16:15
camelia rakudo-moar 278283: OUTPUT«===SORRY!===␤Arg count 2 doesn't equal required operand count 6 for op 'unicmp_s'␤»
cognominal well, we can 6.c.aa 6.c.ab ... 6.c.z at least
samcv m: use nqp; say ORDER(nqp::unicmp_s('v6.c.aa','v6.c.z',7,0,0))
camelia rakudo-moar 278283: OUTPUT«Less␤»
timotimo no matter what you say, we *will* run into that problem at some point :)
ralisch It also has to fit into what distros expect.
samcv also what is the problem
there's lots of numbers
timotimo we have different meanings for letters in the version and numbers in the version
samcv why can't we use hex numbers?
notviki timotimo: which problem? 16:16
samcv or base 32?
like c comes after 9
not immediately but it is higher value imo
timotimo that we run out of the 26 letters :)
notviki doubts
samcv then add another number 16:17
timotimo maybe we can continue with greek letters :P
notviki Or at least.... I'll be dead by then
samcv base 26 numbering system…
timotimo m: say v6.α
camelia rakudo-moar 278283: OUTPUT«v6.α␤»
samcv obviously
timotimo m: say v6.Π
camelia rakudo-moar 278283: OUTPUT«v6.Π␤»
samcv naw
ralisch m: say v6.d̴̸̷̹̬̳̱̰͖̯̘̪̾̎͒̇͗ͩͪͮ̈ͦ͛͂́́͛͝ 16:18
camelia rakudo-moar 278283: OUTPUT«v6.d̴̸̷̹̬̳̱̰͖̯̘̪̾̎͒̇͗ͩͪͮ̈ͦ͛͂́́͛͝␤»
samcv we just go from 6.y to 6.za
^ problem solved guys!!!
AlexDaniel notviki: \o/
ralisch dpkg agrees 6.c.y < 6.c.za
samcv yep 16:19
AlexDaniel yea… interesting solution
please don't forget to mention it in 40 years when this may possibly become relevant
samcv well. it just skips a version
ok 16:20
notviki :P
samcv can it be part of 6.d though
hah
though i guess doesn't really have to be specified. something something
AlexDaniel .in 40 years use 6.c.za after 6.c.y 16:21
yoleaux2 AlexDaniel: Sorry, I don't understand your duration. Try using units: 1h30m, 1d, etc.
AlexDaniel .in 14600d use 6.c.za after 6.c.y
yoleaux2 AlexDaniel: I'll remind you on 23 Dec 2056 11:21Z
AlexDaniel thanks.
samcv what why 6.c
AlexDaniel oh, that's right before Christmas!
samcv implying in 40 years we will still be on 6.c
AlexDaniel oops 16:22
.in 14600d use 6.za after 6.y
yoleaux2 AlexDaniel: I'll remind you on 23 Dec 2056 11:22Z
samcv depressing!
notviki That's a lot of confidence in my ability to run a server :)
AlexDaniel notviki: let's distribute it a little bit…
ok
notviki .in 1460000d use 6.za after 6.y
yoleaux2 notviki: I'll remind you on 8 May 6014 12:22Z
AlexDaniel now THAT is a lot of confidence
2056 is not as unrealistic 16:23
samcv still though
notviki man, ralisch's stuff messed my terminal I can barely see what I'm typing
AlexDaniel notviki: :-/ sounds like you need to fix this 16:24
notviki I guess it's a sign that I should abandon IRC and go study the rakudo internals course
AlexDaniel: it's a pretty beta-looking terminal.
But it's pretty :)
ralisch .in 9223372036854775807d use 7 after 6
yoleaux2 ralisch: I'll remind you on 30 Jul 25252734927768571 12:24Z
notviki Can even use a nyan cat as a selection cursor
hehe: i.imgur.com/KVDkDMA.png 16:27
timotimo that is very important. 16:28
ralisch Sorry for breaking your stuff, I though you were in the better terminal. 16:29
dalek p: c782eb7 | samcv++ | tools/build/MOAR_REVISION:
Bump MoarVM revision to bring in changes

  samcv
   * Implement Bidi_Mirroring_Glyph as an integer property
   * Implement Emoji grapheme breaking and other combined codes as well as
   adding Emoji Unicode properties.
   * Fix breaking after Prepend characters
  jnthn
   * Re-implement UTF8-C8 streaming decode.		 16:31
AlexDaniel “And fear not - your background videos will still play across all the Tabs you have”
yea, that's a very interesting terminal emulator… 16:32
ralisch Doesn't have a scrollbar, but can display thumbnails 16:34
AlexDaniel if only it could display combining characters, bwahahahaha
ralisch xterm also can't but it knows to ignore what it can't handle, instead of blowing up. 16:35
dalek kudo/nom: 7c8b705 | samcv++ | tools/build/NQP_REVISION:
Bump MoarVM revision to bring in changes

  samcv
   * Implement Bidi_Mirroring_Glyph as an integer property
   * Implement Emoji grapheme breaking and other combined codes as well as
   adding Emoji Unicode properties.
   * Fix breaking after Prepend characters
  jnthn
   * Re-implement UTF8-C8 streaming decode.		 16:36
ast: b497b7f | jnthn++ | S32-str/ (2 files):
Move, unfudge, expand UTF8-C8 tests.

Some tests were added in encode.t; they're now in utf8-c8.t so we can just fudge the whole file on JVM, which is a tad easier. The failing tests for RT #128184 are now unfudged, and some new tests are added to provide more complete coverage.
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128184
jnthn \o/
samcv \o/
ralisch 🙌 16:40
samcv adding an RT for this bug too jnthn 16:44
the space one
ralisch If you want a teminal recommendation, here's one: rxvt-unicode can display zalgo text. 16:45
dalek ast: 7574838 | samcv++ | S05-mass/properties-general.t:
TODO failing test, due to a bug in NQP

  rt.perl.org/Ticket/Display.html?id=130483		 16:46
samcv but it doesn't do fall back fonts very well ralisch
unless you know something I don't. also it doesn't resize the fonts to fit the squares 16:47
it will just silently fail to display if the character is too big to fill the pixel instead of scaling it
st is sadly the only terminal that does that properly and fallback fonts, but it doesn't even have scrolling without a patch
so it must not be THAT hard to do. st doesn't have that many lines of code. ugh 16:48
i tried like 12 terminal programs. and that's the only one that resized the characters right AND compose key worked fine
i mostly use Konsole, and then use st when I need to make sure compose works and text won't get improperly vertically alligned with tall characters 16:49
|Tux| jnthn++; # OHHHHHHHHHHHH! He fixed utf8-c8 OHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH What a way to start 2017 16:51
which of course means I have to start working with it
ralisch Fallback fonts are working fine for me in urxvt, but it does use the system for it, so YMMV. It even did Braile, Runes, Thai for me. Amharic failed even though I have the font for it installed. 16:58
notviki AlexDaniel: oh, and this terminal also freezes once in a while so I think no one is talking, even though it's just the terminal that froze 17:02
BUT IT'S PRETTY!
samcv yeah looks like i have to not indent
notviki though I think I saw indented fudges
dalek kudo/nom: b183cab | (Fernando Correa de Oliveira)++ | lib/Test.pm6:
Do not backslash every backslash on test description
 17:03
kudo/nom: 689a768 | (Zoffix Znet)++ | lib/Test.pm6:
Merge pull request #985 from FCO/nom

Do not backslash every backslash on test description
ast: 63af18c | samcv++ | S05-mass/properties-general.t:
Don't indent the TODO so it actually TODO's
 17:04
samcv kk fixed now notviki
u: { .uniprop-int('Bidi_Mirroring_Glyph') } 17:06
unicodable6 samcv, U+0028 LEFT PARENTHESIS [Ps] (()
samcv, U+0029 RIGHT PARENTHESIS [Pe] ())
samcv, U+003C LESS-THAN SIGN [Sm] (<)
samcv, gist.github.com/a4b7c885957c12bf76...363499e695
notviki samcv: still fails and prints "Don't know how to mark statement for TODO!"
samcv works for me
err
maybe need to escape things? 17:07
dalek ast: eebac56 | samcv++ | S05-mass/properties-general.t:
Make TODO lowercase (todo)
 17:08
samcv fixed
pmurias jnthn: do we want the Rakudo setting to depend on arguments with a type constraint being decontainerized?
jnthn: there seems to be a few places where the nqp::decont can be removed if so 17:09
notviki TIL git will outright throw out your commit on git pull --rebase if it pulls a commit that makes same changes as your commit :S
samcv exact same changes? well. 17:10
yeah it does that
hah
it can't do it twice
notviki I expected it to complain
samcv does it warn you though notviki ?
i would too
pmurias throwing out a now noop commits seems reasonable 17:11
notviki nope. Just told me everything's up to date after merge: gist.github.com/zoffixznet/df9de12...f39d96acf4
pmurias: but the commit messages are entirely different.
samcv you still have the commit in the log locally? 17:12
notviki no, totally gone
samcv ah
ralisch Git can't even see any difference between two commits that do the same thing, it can't even trow a warning.
notviki samcv: so what's with that failing test anyway? 17:13
samcv before <:space> matched LF=space 17:14
the LF unicode property
not the space property
notviki And now?
samcv now that space == White_Space
notviki It's just it's part of 6.c-errata. And I see it's failing there too now....
samcv nqp bug where it tries to check property values instead of boolean property names
it was broken before 17:15
notviki OK
samcv in nqp we need to check the property names before the values
bisectable6, use nqp; say nqp::unipropcode('space') == nqp::unipropcode('White_Space') 17:16
bisectable6 samcv, Bisecting by output (old=2015.12 new=689a768) because on both starting points the exit code is 0
notviki star: say " " ~~ /<:space>/
camelia star-m 2016.10: OUTPUT«｢ ｣␤»
bisectable6 samcv, bisect log: gist.github.com/f8b1850fe0735f5763...3213691c65
samcv, (2017-01-02) github.com/rakudo/rakudo/commit/7c...50246afcf0
notviki m: say " " ~~ /<:space>/
camelia rakudo-moar 689a76: OUTPUT«Nil␤»
samcv but if you see the log 17:17
notviki So if all bugs fixed that's supposed to work ^
samcv of bisectable you can see it's been broken for a while
yeah
and so are other LF=space characters in unicode
notviki OK. Then let's leave the test unfudged in 6.c-errata for now.
samcv <ws> works fine
yeah 17:18
m: use nqp; my $var = nqp::unipropcode('Block'); say $var; say nqp::unipvalcode(5, 'Greek') 17:25
camelia rakudo-moar 689a76: OUTPUT«5␤8␤»
samcv m: use nqp; my $var = nqp::unipropcode('space'); say $var; say nqp::unipvalcode(99, 'space')
camelia rakudo-moar 689a76: OUTPUT«99␤0␤»
samcv yeah
it's not properly checking boolean ones. and only ever worked on just LF=space because of the bug 17:26
because space=1 for the space/White_Space property NOT 'space' 17:27
that's basically the gist of it
notviki, 17:32
m: use nqp; my $var = nqp::unipropcode('space'); say $var; say nqp::getuniprop_int(' '.ord, 99)
camelia rakudo-moar 689a76: OUTPUT«99␤1␤»
samcv it should only check unipvalcode for Script and uh. idk maybe one or two more? 17:33
<:Script> can be used bare, but i'm not sure what else. oh and general category
notviki .ask jnthn what's your opinion on this 6.c-errata change: New behaviour adds a space before "#TODO" here: github.com/perl6/roast/blob/6.c-er...tput.t#L46 in exchange for escaping only "#" and not every backslash. Can we make this change to 6.c-errata? 17:37
yoleaux2 notviki: I'll pass your message to jnthn.
notviki .ask jnthn the change being changing the test to look for any amount of space before "#TODO" rather than zero 17:38
yoleaux2 notviki: I'll pass your message to jnthn.
jnthn |Tux|: I started re-doing it at the end of last year for the non-streaming case, and got the streaming case in today so it could be merged. :) Hope it helps. 17:39
yoleaux2 17:37Z <notviki> jnthn: what's your opinion on this 6.c-errata change: New behaviour adds a space before "#TODO" here: github.com/perl6/roast/blob/6.c-er...tput.t#L46 in exchange for escaping only "#" and not every backslash. Can we make this change to 6.c-errata?
17:38Z <notviki> jnthn: the change being changing the test to look for any amount of space before "#TODO" rather than zero
jnthn pmurias: Well, if we're going to promise the behavior across backends I guess we can. How difficult is it for you to provide it in rakudo-js? 17:40
samcv ok i gotta go to bed. night all
jnthn 'night, samcv 17:41
pmurias samcv: g'night
jnthn: shouldn't be a problem
jnthn notviki: I assume both are valid by the TAP spec? 17:42
pmurias jnthn: I think I'll switch to using the lower_signature like moarvm does as it seems faster 17:43
jnthn: and I'll have to do it anyway as the setting already depends on that behavior 17:44
notviki jnthn: it doesn't say anything about preceeding space, but if anything the new behaviour is more correct, because the space we added is present in the examples they show: testanything.org/tap-version-13-spe...todo-tests
jnthn pmurias: Yeah, implementing lower_signature will be a big win 17:45
notviki: OK, don't have any objections
pmurias why doesn't the JVM us the lower_signature?
notviki Thanks.
jnthn pmurias: Never got around to implementing the missing bits so it can do so 17:46
pmurias: Combined with the fact it'll increase code size 17:47
dalek ast: 3050638 | (Zoffix Znet)++ | S24-testing/3-output.t:
Relax whitespace requirement before "#TODO" in test output

Needed by github.com/rakudo/rakudo/commit/b1...86782ee788 Change approved by jnthn: irclog.perlgeek.de/perl6-dev/2017-...i_13838067		 17:49
pmurias should I make the nqp-level Binder.bind_sig do the decontainerization too? as it changes the semantics a tiny bit
jnthn pmurias: That's probably fine
dalek ast/6.c-errata: c44a181 | (Zoffix Znet)++ | S24-testing/3-output.t:
Relax whitespace requirement before "#TODO" in test output

Needed by github.com/rakudo/rakudo/commit/b1...86782ee788 Change approved by jnthn: irclog.perlgeek.de/perl6-dev/2017-...i_13838067		 17:50
jnthn m: use Test; ok qq["\c[COMBINING TILDE]"] ~~ / ^ :ignoremark \"/ 18:01
camelia rakudo-moar 689a76: OUTPUT«not ok 1 - ␤␤# Failed test at <tmp> line 1␤»
dalek p: 858b946 | jnthn++ | src/QRegex/P6Regex/Actions.nqp:
Change rxtype of a literal to use literal.

Not quite sure why it used `enumcharlist`; it would work, but the generated code would be lower quality.		 18:03
p: 1a51a91 | jnthn++ | src/QRegex/P6Regex/Actions.nqp:
Fix ignoremark on backslashed literals.
lizmat hmmm.... power outage :-( 18:06
dalek kudo/nom: 6188771 | jnthn++ | tools/build/NQP_REVISION:
Bump NQP_REVISION for regex fix.

Fixes compilation of `/:ignoremark \"/` to actually ignore marks.		 18:07
ast: e0bd1e6 | jnthn++ | S05-modifier/ignoremark.t:
Test to cover RT #130465.
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130465
jnthn read it as "power outrage" somehow :) 18:54
dugword PR for new roast tests added for resolved RT issues: github.com/perl6/roast/pull/212 Please let me know if I need to make any changes. 19:09
notviki dugword: are those the only possible families; ever? I see there are a couple more constants defined: github.com/rakudo/rakudo/blob/nom/...T.pm#L3-L7 19:40
... also makes me wonder if the new :$famility type constraint is too restrictive. 19:41
geekosaur only families with a decent chance of being available everywhere 19:43
yoleaux2 31 Dec 2016 14:47Z <notviki> geekosaur: what were the problems with just using 1/0, -1/0, 0/0 for roundtripping? irclog.perlgeek.de/perl6-dev/2016-...i_13826225
geekosaur ...I do not recall, I only recall it being one of those Zefrem tickets 19:44
notviki Zefram.
geekosaur but possibly related to the fact that NaNs carry (implementation dependent) information about the kind of NaN and what generated it
(Inf does not) 19:45
notviki pmurias: are you around? A while back you gave me code to make my hand rolled compiler take script filenames and run them. It was something about making @*ARGS available in the program. Do you remember what that code was? This is my current compiler I'm trying to make take script files: gist.github.com/zoffixznet/c06ecaf...94523f5288 19:46
geekosaur btw I still worry about hardcoding those packet families etc., I don't think there's any guarantee of them being the same everywhere (although there is certainly a tendency to use the standard ones because of people hardcoding values and expecting it to work...) 19:47
pmurias notviki: looking... 19:53
notviki oh damn, I forgot the .new on QAST::Var 19:55
pmurias: thanks. This works now: gist.github.com/zoffixznet/59cc8f7...dab542a668
I forgot .new() on Var and also needed :decl<param> 19:56
notviki doesn't see any of Zefram's tickets for "Rat" or "roundtriping" mentioning anything for Rat Inf/NaN roundtripping 20:00
lizmat jnthn: yeah, we had about 45 mins of outage, and after that one of our servers didn't automatically reboot and the port forwarding on a router had been shot to bits 20:03
all ok again now
dugword notviki: A good point, but would PF_UNIX (Unix sockets) be an INET socket? Perl5 has separate modules for INET and UNIX 20:06
I can also change it to look for $family <= PF_MAX since that implies it should be the max value? 20:07
notviki I don't know. But the constants are in there and currently the test says that all Perl 6 implementations must fail if they detect a family that's neither 2 nor 3. Perhaps that test can be relaxed a bit on what values it should die for 20:08
geekosaur probably not a good idea to conflate AF_LOCAL and AF_INET
notviki m: PIO::PF_INET
camelia rakudo-moar 618877: OUTPUT«Could not find symbol '&PF_INET'␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
geekosaur thinks the test should verify that 2 and 3 are supported, and not try to dictate what other values might be supported 20:09
notviki I don't get why those constants aren't available to the user
dugword Okay, I have to leave for a while, but I can update the test. I'll also have to update the IO::Socket::INET code to allow for values in addition to 2 and 3? Should it have constraints to the constants, or just no constraints? 20:10
notviki has no idea 20:11
dugword I'll look around the code tonight and see if PF_LOCAL or PF_UNIX are ever used, and do some research to refresh on networking to see what would be appropriate 20:12
I think though that we are decades away from a IPv7+, considering the state of IPv6. We could probably deal with a new IP version when it happens :) 20:14
dogbert2 m: say ProtocolFamily(PF_INET).value 20:43
camelia rakudo-moar 618877: OUTPUT«2␤»
notviki Thanks
jnthn lizmat: Glad it's all working again :) 20:44
notviki m: say PF_INET
camelia rakudo-moar 618877: OUTPUT«PF_INET␤»
notviki Actually, I still don't get. 20:48
We have a duplicated set of those constants? one in IO.pm and another in INET.pm
dogbert2 anyone knows what this means (strace after connecting to hanging process): futex(0xae5070c, FUTEX_WAIT_PRIVATE, 9, NULL 21:05
it's what I get when t/spec/S32-io/socket-accept-and-working-threads.t hangs which it does once in a while 21:06
lizmat jnthn: are the utf8-c8 in rakudo HEAD now ? 21:09
*fixes
jnthn: nevermind, it's part of 7c8b705b3375522fc 21:10
RabidGravy dogbert2, it means that most probably there is some threading deadlock 21:16
dogbert2 RabidGravy: thx, something is definitely going horribly wrong 21:19
geekosaur sadly, threading issues are not exactly new :/ 21:23
lizmat samcv jnthn: am I correct in seeing that nqp::unicmp_s is not yet exposed in Perl 6 ? 21:31
dalek kudo/js: 2da8192 | (Pawel Murias)++ | src/vm/js/Perl6/Ops.nqp:
[js] Fix comment.
 21:44
rakudo/js: ee7cd70 | (Pawel Murias)++ | src/vm/js/Perl6/Ops.nqp:
rakudo/js: [js] :sideffects -> :side-effects
ast: a14cb3d | (Douglas Jenkins)++ | S32-io/socket-fail-invalid-values.t:
Added test to fail on invalid port or family values
 22:21
ast: 4a45af0 | lizmat++ | S32-io/socket-fail-invalid-values.t:
Merge pull request #212 from dugword/bugfix-RT-130473-130475/fail-invalid-values

Added test to fail on invalid port or family values
kudo/nom: 6dd2995 | (Douglas Jenkins)++ | t/spectest.data:
Added new test entry for socket-fail-invalid-values.t
kudo/nom: a566efb | lizmat++ | t/spectest.data:
Merge pull request #986 from dugword/new-test/socket-fail-invalid-values

Added new test entry for socket-fail-invalid-values.t
dugword Woo! 22:23
lizmat and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2017/01/02/...rime-time/ 22:27
notviki lizmat++ 22:30
Yey, first weekly of the week \o/
lizmat week? 22:31
notviki 150 commits in a week :o
Err, I mean year :D
lizmat :-)
notviki m: say "News" if True but (rand > .5) 22:35
camelia rakudo-moar a566ef: OUTPUT«News␤»
samcv lizmat++ 22:49
lizmat, She implemented the Unicode Collation Algorithm in NQP this should say in MoarVM 22:51
well i mean it has a nqp op but. idk
nqp doesn't have it for all backends though
dalek ast: a0dbb3b | lizmat++ | S15- (2 files):
Unfudge now passing tests, samcv++
samcv yay!
dugword Hey I'm on the news letter :)
lizmat samcv: fixed 22:52
samcv: changed to MoarVM/nqp 22:53
samcv sweet
jnthn lizmat++ # weekly 22:58
'night, #perl6-dev 23:24
samcv uhm seeing this test in roast 23:43
m: use Test; lives-ok { EVAL "q\x298d test \x298e" }, "Unicode open-298d maps to close-298e";
camelia rakudo-moar a566ef: OUTPUT«ok 1 - Unicode open-298d maps to close-298e␤»
samcv but it doesn't
they're not mirroring glyphs and i believe not tested anywhere else. should be ok to change it?
unless it's testing that it doesn't die not that it works? not sure???
m: say "q\x298d test \x298e" 23:44
camelia rakudo-moar a566ef: OUTPUT«q⦍ test ⦎␤»
samcv m: say EVAL "q\x298d test \x298e"
camelia rakudo-moar a566ef: OUTPUT« test ␤»
samcv m: '⦍'.uniprop('Bidi_Mirroring_Glyph').ord.base(16).say
camelia rakudo-moar a566ef: OUTPUT«2990␤»
samcv it has a mirroring glyph. but the tested one is not it 23:45
samcv hopes it's not in 6.c-errata
with my change it looks for the _actual_ matching bracket not the one we have programmed in ;P 23:46
well proposed changed to nqp
i think if we _have_ exceptions to "only allow Bidi mirroing _glyphs" + some extras 23:48
those 'some extras' should NOT include characters which DO have matching glyphs
m: say Q<｢>.uniprop('Bidi_Mirroring_Glyph') 23:49
camelia rakudo-moar a566ef: OUTPUT«｣␤»
samcv for example this one is fine 23:50
m: "\x[298d]\x[2990]".uninames.say 23:51
camelia rakudo-moar a566ef: OUTPUT«(LEFT SQUARE BRACKET WITH TICK IN TOP CORNER RIGHT SQUARE BRACKET WITH TICK IN TOP CORNER)␤»
samcv weeps. sees that is in 6.c-errata :((( 23:54
that is depressing
Couldn't find terminator ⦐ (corresponding ⦍ was at line 33) 23:55

Powered by Cro and the Raku® Programming Language.

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

Thank you!