dalek | kudo/nom: aada71a | TimToady++ | src/Perl6/World.nqp: want nibble_to_str's ast |
00:31 | |
kudo/nom: 5334cb7 | TimToady++ | src/Perl6/Actions.nqp: propagate sinkness to thunks of and/or etc. (Note, apparently unhides a bug in roles where generic statements are not correctly marked as wanted, or incorrectly sunk.) |
04:41 | ||
TimToady | m: say [1,2,3] or 42 # will warn soon | 04:42 | |
yoleaux2 | 26 Jul 2016 02:24Z <gfldex> TimToady: i somehow feel that definedness contraints on --> and returns are to missleading to have. see: m: class Baddy is Nil {}; sub niler( --> Any:D ) { Baddy.new }; my $i = 42; $i = niler; dd $i; | ||
camelia | rakudo-moar aada71: OUTPUT«[1 2 3]» | ||
yoleaux2 | 26 Jul 2016 02:32Z <gfldex> TimToady: what makes me uneasy are lines line the following. A new and aspiring Perl 6 Programmer might take the `--> IO::Path:D` at face value. Given it's IO and we support Windows I would not bet my life on the :D part. docs.perl6.org/type/IO$COLON$COLON...d_absolute | ||
TimToady | m: 42 with [1,2,3] | 04:44 | |
camelia | ( no output ) | ||
TimToady | m: 42 ?? 43 !! 44 | ||
camelia | ( no output ) | ||
dalek | ast: 7345b3b | usev6++ | S04-declarations/smiley.t: Fudge tests for RT #127958 for rakudo-j |
04:45 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=127958 | ||
TimToady | m: say [1,2,3] or 42 | 04:52 | |
camelia | rakudo-moar 5334cb: OUTPUT«WARNINGS for <tmp>:Useless use of constant integer 42 in sink context (line 1)[1 2 3]» | ||
TimToady | m: 42 with [1,2,3] | ||
camelia | rakudo-moar 5334cb: OUTPUT«WARNINGS for <tmp>:Useless use of constant integer 42 in sink context (line 1)» | ||
TimToady | m: 42 ?? 43 !! 44 | ||
camelia | rakudo-moar 5334cb: OUTPUT«WARNINGS for <tmp>:Useless use of constant integer 43 in sink context (line 1)Useless use of constant integer 44 in sink context (line 1)» | ||
TimToady | m: 43 xx 44 | ||
camelia | ( no output ) | ||
TimToady | dunno why that one doesn't warn yet... | ||
m: my $x = 42; $x xx 44 | 04:54 | ||
camelia | rakudo-moar 5334cb: OUTPUT«WARNINGS for <tmp>:Useless use of $x in sink context (line 1)» | ||
TimToady | at least we get that much | ||
m: my $x = 42; $x++ xx 44 | 04:55 | ||
camelia | ( no output ) | ||
lizmat | good *, #perl6-dev | 07:42 | |
lizmat gets a warning during stage optimize now | 07:43 | ||
in Enumeration: multi method pick(::?CLASS:D: *@pos) { self xx +?( @pos[0] // 1 ) } | 07:46 | ||
and: multi method roll(::?CLASS:D: *@pos) { self xx +?( @pos[0] // 1 ) } | 07:47 | ||
lizmat wonders if TimToady's last change uncovered a bug | |||
jnthn | lizmat: Well, TimToady did note it uncovers a bug, in the commit message :) | 08:45 | |
nine | The code in question doesn't look all that buggy to me. It also seems to be spectested | 08:53 | |
or maybe not | 08:56 | ||
Actually I don't quite understand what those multi candidates are supposed to do. | 09:19 | ||
lizmat: What's multi method pick(::?CLASS:D: *@pos) for? It only uses the first value in @pos anyway. | 09:23 | ||
AFAICT the warning is wrong. The code works as is. Putting a return before the self fixes the warning and doesn't change behavior, so it looks like the sink warning is failing on the implicit return. | 09:24 | ||
jnthn | Righty, time to do Perl 6 stuff. :) | 09:33 | |
rt.perl.org/Ticket/Display.html?id=128736 seems a little concerning... | 09:34 | ||
lizmat | jnthn: I consider #128736 to be on my plate, since I probably borked it in the first place | 11:02 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128736 | ||
lizmat | so, feel free to pursue, or to leave it to me :-) | 11:03 | |
jnthn | I've hunting a GC orchestration hang and examining its performance today, which I think will keep me more than busy enough :) | 11:32 | |
lizmat | ok :-) | 11:38 | |
jnthn | One of those fun jobs where different profilers can give you results a factor of 100 different :) | 11:39 | |
lizmat | yuck :-( | ||
jnthn | Well, just means you gotta understand their methodology and which one to care about :) | 11:40 | |
jnthn breaks out a third profiler :) | 11:50 | ||
[Coke] | I did a little debug on RT #125501 - could use an actual windows developer to peek at it. (for one, it didn't seem to create the install/ directory before it needed it) | 13:06 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=125501 | ||
dalek | kudo/nom: 6d1e95c | lizmat++ | src/core/Rakudo/Internals.pm: Remove .plan It was a plan once, but hasn't been for a while If we ever decide to revive the .plan idea, we can add it again. |
13:29 | |
p: 94712df | (Pawel Murias)++ | src/vm/js/Operations.nqp: [js] Remove old TODO. |
13:43 | ||
p: 5647d54 | (Pawel Murias)++ | src/vm/js/nqp-runtime/sixmodel.js: [js] Remove old TODOs. |
|||
TimToady | my guess is that the uncovered bug has something to do with the method being in a role somehow, and maybe compiling generic code doesn't quite do the wanting of the final expression somehow | ||
but it's a benign warning, albeit incorrect | 13:44 | ||
datapoint: putting in an explicit return suppresses the warning | 13:48 | ||
(because return explicitly wants its argument) | 13:49 | ||
m: $_ = 'datapoint'; say S:5th/''/ / | 13:50 | ||
camelia | rakudo-moar 5334cb: OUTPUT«data point» | ||
TimToady | no, I'm wrong about generics | 13:56 | |
m: my method bar() { self xx 42 } | |||
camelia | rakudo-moar 6d1e95: OUTPUT«WARNINGS for <tmp>:Useless use of self symbol in sink context (line 1)» | ||
TimToady | m: my method bar() { pi xx 42 } | ||
camelia | ( no output ) | ||
TimToady | seems to be a difficulty with self, perhaps can't take the WANTED annotation | 13:57 | |
well, that should be pretty easy to chase down after I actually wake up | 13:58 | ||
m: my method bar($a) { $a xx 42 } | 14:02 | ||
camelia | rakudo-moar 6d1e95: OUTPUT«WARNINGS for <tmp>:Useless use of $a in sink context (line 1)» | ||
TimToady | no, it's pi that is too simple to take an annotation | 14:03 | |
(iirc) | |||
ah, no, xx optimizes constants differently, should really wake up before spamming the channel... | 14:05 | ||
[Coke] | zoffix, RT#128744 - I assume you meant NYI, not NIY? | 14:06 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128744 | ||
TimToady | Not Implemented Yet means the same thing:) | ||
[Coke] | but no one searching for NYI will find it. :| | ||
[Coke] changes it | |||
lizmat | masak: my @a = (); @a.splice(0,0,42) # workaround for RT #128736 | ||
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128736 | ||
TimToady | m: $_ xx 1 # minimal test case is minimal | 14:07 | |
camelia | rakudo-moar 6d1e95: OUTPUT«WARNINGS for <tmp>:Useless use of $_ in sink context (line 1)» | ||
TimToady | well, except that *is* in sink context | ||
m: $_ = -> $a { $a xx 1 } | 14:08 | ||
camelia | rakudo-moar 6d1e95: OUTPUT«WARNINGS for <tmp>:Useless use of $a in sink context (line 1)» | ||
TimToady | and that shouldn't be sank | 14:09 | |
unmatched} | [Coke]: yeah | 14:11 | |
I don't know why people use NYI.. It sounds Engrish to put "Yet" in the middle | |||
lizmat | I guess that's one of the few differences between English and American | 14:12 | |
unmatched} | Is it common to put it in the middle in UK? | ||
lizmat | it has more to do with French grammar influences in English, as opposed to the German grammar influences in American | 14:13 | |
actually, thinking about it, NYI is more American | 14:14 | ||
Not Implemented Yet would be more French "pas implemente encore" or something like that :-) | |||
whereas german: noch nicht implementiert aka not yet implemented | 14:15 | ||
unmatched} | That may be why NIY sounds more pleasant to my Canadian ear | ||
lizmat | but who am I as a native Dutch speaker :-) | ||
DrForr | Of course the acronym would need to be a further permutation,maybe YIN. | ||
MasterDuke | English is a Germanic language | 14:16 | |
geekosaur | I, nd most people I know, would say "not implemented yet". I thought the ordering was a matter of emphasis, making sure the "yet" was not dangling off the end letting the "not implemented" be primary | ||
MasterDuke, but it is the result of a trainwreck between two Germanic languages and a French language (iirc old Normandy's French was somewhat distinct from the Parisian French that became "the standard") | 14:18 | ||
MasterDuke | true, i wouldn't say it took any of the regularness of any of its parent languages | 14:19 | |
TimToady | no, German would translate to YNI | 14:20 | |
geekosaur | ...that's what I thought... | 14:21 | |
unmatched} | m: "NIY".comb.permutations».join.say | 14:22 | |
camelia | rakudo-moar 6d1e95: OUTPUT«(NIY NYI INY IYN YNI YIN)» | ||
unmatched} | \o/ | ||
MasterDuke | which in English could be 'as yet not implemented' | ||
or 'as of yet not implemented' | 14:23 | ||
dalek | ast: 0a3a0e9 | lizmat++ | S09-multidim/methods.t: There is no .plan for the foreseeable future |
15:11 | |
ast/6.c-errata: f64a136 | lizmat++ | S09-multidim/methods.t: There is no .plan for the foreseeable future |
15:12 | ||
kudo/nom: 3bea9d8 | lizmat++ | src/core/Rakudo/Internals.pm: Make sure that .pop|shift|splice fail here For some reason, other candidates of .splice were leaking through here, causing spectest failure with the wrong exception. |
15:14 | ||
lizmat | afk& | 15:16 | |
[Coke] | should NQP tickets go into github nqp issues or is rt perl6 queue fine? | 15:27 | |
because we have them in both places right now. | |||
I am OK with having multiple bug queues, rather than trying to categorize everything in RT. | 15:28 | ||
anyone else have a preference? | |||
unmatched} | I like Github over RT, but I'm ok with having both queues if only to not spend labour on moving tickets from one queue to another | 15:30 | |
jnthn | [Coke]: Github issues for NQP | ||
jnthn almost dealt with that earlier then got distracted hutning concurrency bugs :) | 15:31 | ||
.oO( If you hunt this circular wait, I'll move the ticket? ;) ) |
|||
unmatched} | :) | ||
[Coke] | jnthn: that's where I'm leaning, as well. I'll migrate any tickets I find there, and try to update github.com/rakudo/rakudo/wiki to have an explanation of our list of places to report bugs. | 15:32 | |
dalek | kudo/nom: 948aee0 | jnthn++ | src/core/Channel.pm: Re-order code to avoid a race condition. This could lead to deadlocks, such as a hang in S17-supply/syntax.t on the test involving multiple whenevers over multiple channels and Promises synchronizing order. |
15:35 | |
[Coke] wonders if this really old nqp ticket is still relevant: github.com/perl6/nqp/issues/145 | 15:37 | ||
timotimo | fantastic! | ||
jnthn | [Coke]: Yeah, we should still get around to that at some point :) | 15:39 | |
Not a priority (quite evidently ;)) | |||
But worthwhile | |||
[Coke] finds a perl6 ticket where the bug report is on: | 15:43 | ||
p6: use nqp; say nqp::index("xxy","y",-1) | 15:44 | ||
camelia | rakudo-moar 3bea9d: OUTPUT«-1» | ||
[Coke] | aw, no jvm? (or is it just that slow?) | ||
unmatched} | r: use nqp; say nqp::index("xxy","y",-1) | 15:46 | |
camelia | rakudo-moar 3bea9d: OUTPUT«-1» | ||
..rakudo-jvm cd19db: OUTPUT«2» | |||
unmatched} | p6 is same as m: github.com/perl6/evalbot/blob/mast...bot.pl#L70 | 15:47 | |
[Coke] | ack, someone moved things around! :) | 15:49 | |
ok, moved the few nqp-only tickets I could find. | 15:52 | ||
unmatched} | I think I'm out of steam with RT#126119 | 15:57 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=126119 | ||
unmatched} | I don't think what doing fancy math trying to guestimate future leapseconds serves other than to scratch some OCD itch so you could say "it's less wrong now" | 15:58 | |
s/think/know/; | |||
unmatched} points to the a response on -compiler list that isn't on RT ticket [yet] | 16:00 | ||
I wonder what Zefram is working on with P6. Their tickets are always about these rare super-edge case scenarios like future gigasecond leap seconds and decorators on numbers | 16:04 | ||
TimToady | he just has a really good feel for boundaries, I think | 16:05 | |
mst | he likes doing things that should theoretically work and seeing what happens | 16:10 | |
unmatched}: I'd note that the fact that I also like this got me an honorary Moose core team membership for being a walking destruction test, and at least somewhat contributed to perl6 not ending up with the same edge cases | 16:11 | ||
unmatched} | To note: I'm not complaining. It's good this is being done. I was just curious on what leads to discovery of such bugs :) | 16:12 | |
m: "q[\x[308]foo]\x[308]".EVAL.perl # other one from Zefram | |||
camelia | rakudo-moar 948aee: OUTPUT«===SORRY!=== Error while compiling /home/camelia/EVAL_0Couldn't find terminator [̈ (corresponding [̈ was at line 1)at /home/camelia/EVAL_0:1------> q[̈foo]̈⏏<EOL> expecting any of: [̈» | ||
mst | so ... I think really his point with that one is "if you're going to guess, here's how to guess right" | ||
unmatched} | m: say "[\x[308]" ~~ /^<:Ps>$/ | 16:13 | |
camelia | rakudo-moar 948aee: OUTPUT«「[̈」» | ||
mst | I think he considers "actually, fuckit, don't guess" to be obviously an option, so didn't explicitly mention it | ||
unmatched} | Would be neat to see some actual algorithm for guestimating leap seconds to see their performance penalty and how accurate and most-importantly predictable they are. | 16:14 | |
m: DateTime.new("2016-12-31T23:59:60") | 16:15 | ||
camelia | ( no output ) | ||
unmatched} | star: DateTime.new("2016-12-31T23:59:60") | ||
camelia | star-m 2016.04: OUTPUT«Second out of range. Is: 60, should be in 0..^60; There is no leap second on UTC 2016-12-31 in block <unit> at <tmp> line 1» | ||
mst | unmatched}: if you asked, I'm sure Zefram could provide you more information than you ever wanted ;) | 16:16 | |
unmatched} | :) | 16:17 | |
arnsholt | [Coke]: I started an attempt at killing of O being weird at some point, but foundered on some issue or other | 16:21 | |
You get into bootstrappy issues. The stage0 compiler expects the current API, so it can't just be replaced wholesale | 16:22 | ||
Ah, except no. As I point out in a comment on the issue =D | 16:33 | ||
jnthn | m: (supply { whenever Supply.from-list(1..10) { my $val = $_; LAST { say $val; } } }).tap(&say) | 16:47 | |
camelia | rakudo-moar 948aee: OUTPUT«10» | ||
jnthn | m: (supply { whenever Supply.from-list(1..10) { my $val = $_; LAST say $val; } }).tap(&say) | ||
dalek | kudo/nom: 4c848b0 | TimToady++ | src/Perl6/Actions.nqp: make sure xx propagates wantedness correctlier We needed to handle the thunky and non-thunky cases differently. (Also, methods were not marking their final statement as wanted.) |
||
camelia | rakudo-moar 948aee: OUTPUT«(Any)» | ||
jnthn | bisectable: (supply { whenever Supply.from-list(1..10) { my $val = $_; LAST say $val; } }).tap(&say) | 16:48 | |
bisectable | jnthn: On both starting points the exit code is 0 and the output is identical as well | ||
jnthn: Output on both points: (Any) | |||
jnthn | star: (supply { whenever Supply.from-list(1..10) { my $val = $_; LAST say $val; } }).tap(&say) | ||
camelia | star-m 2016.04: OUTPUT«(Any)» | ||
jnthn | Hmm | ||
How the heck did Supply.tail work before with that bust... | 16:49 | ||
(It uses it and very mysteriously works o.O) | 16:50 | ||
Or rather, worked before I fixed another broken thing | |||
TimToady | with latest change, finds a real bug in the setting | 16:54 | |
in src/core/Block.pm: my $need_cap = $sig.count == Inf and not ($slurp_p and $slurp_n); | |||
skids: ^^^ your code, according to git :) | |||
buy maybe you just reindented someone else's code... | 16:56 | ||
skids | So...what was the latest change? | 16:58 | |
TimToady | gets 'Useless use of "not"' with my latest wantedness checks | 16:59 | |
because 'and' is looser than = | |||
skids | Oh. Yeah that's a problem. | 17:00 | |
Hrm not sure what the actual affect might be. | 17:03 | ||
*effect | |||
I think in the misbehaving case it just runs an unecessary grep and a couple unecessary ||=s | 17:08 | ||
[Coke] | .u O | 17:09 | |
yoleaux2 | U+004F LATIN CAPITAL LETTER O [Lu] (O) | ||
[Coke] | RT: 1325; @LARRY: 4; CONC: 8; GLR: 6; JVM: 66; LHF: 1; LTA: 83; NEW: 864; NYI: 30; OSX: 6; PERF: 18; POD: 9; PRECOMP: 8; RFC: 24; SEGV: 27; STAR: 1; TESTNEEDED: 10; TODO: 8; UNI: 12; UNTAGGED: 650; WEIRD: 3 | 17:10 | |
TimToady: @LARRY tickets might be of interest to you. | 17:11 | ||
unmatched} | To be fair—as originally planned—@LARRY isn't just for Larry, but for all core team. | 17:12 | |
m: (1,2,3)[1] := 4 | 17:13 | ||
camelia | ( no output ) | ||
unmatched} | m: (1,2,3)[1] := 4; say "wut" | ||
camelia | rakudo-moar 4c848b: OUTPUT«wut» | ||
TimToady wonders whether he's @LARRY[0] or @LARRY[*-1]... | |||
unmatched} | m: dd (1,2,3)[1] := 4; | ||
camelia | rakudo-moar 4c848b: OUTPUT«Failure.new(exception => X::Bind.new(target => Any), backtrace => Backtrace.new)» | ||
unmatched} | TimToady: you're @LARRY.pick :) | ||
TimToady | don't be so picky! | 17:14 | |
unmatched} | m: sink ((1,2,3)[1] := 4) | ||
camelia | ( no output ) | ||
[Coke] changes some [spec] tickets to [@LARRY] tickets. | 17:15 | ||
also some STD tickets. | 17:16 | ||
unmatched} | m: my $list = (1,2,3); $list[1] := 4; dd $list | ||
camelia | rakudo-moar 4c848b: OUTPUT«List $list = $(1, 2, 3)» | ||
unmatched} | bisect: my $list = (1,2,3); $list[1] := 4; dd $list | 17:17 | |
bisectable | unmatched}: Exit code on a 'good' revision is 1 (which is bad), bisecting with inverted logic | 17:18 | |
unmatched}: bisect log: gist.github.com/50508630b16fc42867...1f24766a42 | |||
unmatched}: (2016-07-09) github.com/rakudo/rakudo/commit/959cd3b | |||
dalek | kudo/supply-circular-wait-fix: a1ed2c9 | jnthn++ | src/core/Supply.pm: Eliminate locking in favor of a work queue. The approach taken so far is vulnerable to deadlock, since there can be unfortunate timing interactions between the up-stream management of the subscription and downstream flow of values. This changes puts a work queue in place instead, which successfully avoids the deadlock and in theory should give lower overhead, less blocked threads, and hopefully better CPU cache behavior. Regresses a couple of tests, however. |
17:20 | |
unmatched} | RTed: rt.perl.org/Ticket/Display.html?id=128755 | 17:21 | |
dalek | kudo/supply-circular-wait-fix: 7b554a5 | jnthn++ | src/Perl6/Actions.nqp: Fix a scoping bug with blockless phasers. This in turn fixes a regression in S17-supply/tail.t, however it's unclear how it managed to work in the past. |
||
jnthn | throttle.t is still bust, but I'm getting a bit tired for figuring that one out now. | 17:23 | |
(It's only bust in the control messages case, curiously.) | |||
[Coke] | jnthn: tagging RT #125460 as [CONC] (an old S17 ticket) | 17:25 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=125460 | ||
[Coke] | m: use QRegex:from<NQP>; | 17:26 | |
camelia | ( no output ) | ||
[Coke] | m: use NQPP6QRegex:from<NQP>; | 17:27 | |
camelia | ( no output ) | ||
[Coke] | m: use NQPP5QRegex:from<NQP>; | ||
camelia | ( no output ) | ||
jnthn | [Coke]: OK, thanks...good to have things tagged up :) | 17:28 | |
afk for a bit | 17:29 | ||
[Coke] | rt.perl.org/Ticket/Display.html?id=120443 - this looks closable, anyone disagree? | ||
m: class A is CORE::Int {} ; RT #126384 | 17:36 | ||
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=126384 | ||
camelia | rakudo-moar 4c848b: OUTPUT«===SORRY!===Could not locate compile-time value for symbol CORE::Int» | ||
unmatched} | m: ; RT | 17:37 | |
camelia | rakudo-moar 4c848b: OUTPUT«===SORRY!=== Error while compiling <tmp>Undeclared name: RT used at line 1» | ||
[Coke] | lizmat: you last touche rt.perl.org/Ticket/Display.html?id=126664 - is that closable now? | ||
TimToady: there are bunch of rts mentioning "sink" if you're poking around there. :) | 17:40 | ||
unmatched}: looks like a config issue might be impacting mail traffic to RT. the bugadmins are on it. | 17:48 | ||
we have a choice to "reinject missed messages" - it will likely cause a bunch of confusion if this happens; my take is that take any missed messages as missed, and move on. Any thoughts? | 17:49 | ||
unmatched} | Yeah. Anything missed is missed and move on. | 17:51 | |
[Coke] | I'm also volunteering to become one of the human spam monitors. | 17:53 | |
unmatched} imagines www.blogherald.com/wp-content/uploa...00x273.jpg | 17:55 | ||
perlpilot | unmatched}: Not "human spam" monitor, but human "spam monitor" :) | 18:11 | |
unmatched} | :) | ||
[Coke] | jnthn: found another segfault for you in RT #123434 | 18:15 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=123434 | ||
japhb | It's also a matter of what gets emphasized: NIY => Not Implemented YET (but will be soon), NYI => Not Yet IMPLEMENTED (but discussed or designed) | 18:18 | |
Gah, stupid scrolling bug (thought I was current, but just replied to an hours-old conversation) | 18:19 | ||
TimToady | sorry, that conversation is NIA | 18:20 | |
dalek | p/better-O: 3105e6a | arnsholt++ | src/ (2 files): Make HLL::Grammar.O take named arguments. This fixes the problem of the method manually parsing a string (#145). Since the old parser code allowed a construct like '%additive, :op<add>' to copy values from %additive and set the 'op' key to be 'add', I've added a key 'inherit' which starts out with the named information and then writes in the named arguments. |
18:25 | |
arnsholt | So, fixing #145 turned out to be eminently doable | ||
It changes the API a bit though, so if people have time for a brief review/critique, that'd be grand | 18:26 | ||
[TuxCM] | lizmat, current build fails on Text::CSV | 18:28 | |
lizmat | [Coke]: probably needs tests still | 18:30 | |
japhb | [TuxCM]: There have also been guts-ish commits by TimToady and jnthn | 18:31 | |
(Depending on when your last successful build was) | |||
lizmat | [TuxCM]: builds ok for me, but has problems in the .Supply tests | 18:35 | |
TimToady | my changes should only have affected warnings | ||
yes, supplies are busted till we resync with jnthn's branches, I suspect | |||
lizmat is about to be afk again | 18:36 | ||
so I'm not a person to be doing that right now | |||
TimToady | I think jnthn has to decide when to merge back | ||
dalek | kudo/nom: debc7c2 | lizmat++ | src/core/List.pm: Fix for RT #128755 |
18:41 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128755 | ||
lizmat | afk& | ||
[Coke] | down to 598 untagged tickets. | 18:49 | |
dalek | ast: 4e09c41 | (Zoffix Znet)++ | S03-binding/nonsense.t: Fudge failing and hanging tests |
18:56 | |
ast: 27bee13 | (Zoffix Znet)++ | S03-binding/nonsense.t: Binding into List item throws RT#128755 |
|||
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128755 | ||
kudo/nom: f0abe86 | (Zoffix Znet)++ | t/spectest.data: Include S03-binding/nonsense.t in list of run tests -The failing/haning tests have now been fudged. -The file is not present in 6.c-errata branch |
18:59 | ||
[Coke] | rt.perl.org/Ticket/Display.html?id=124571 - Pretty sure we snuck lazy in, didn't we? | 19:00 | |
ah, no, tests still todo'd. | 19:01 | ||
unmatched} | Then maybe they need to be un-todoed | ||
m: my @x = lazy 1..Inf; say @x[42] | 19:02 | ||
camelia | rakudo-moar debc7c: OUTPUT«43» | ||
[Coke] | ... if they're todo'd, we'd be seeing them pass unexpectedly if they worked. | ||
unmatched} | Oh, right | ||
m: my $x = lazy 1..3; say $x.is-lazy | |||
camelia | rakudo-moar debc7c: OUTPUT«True» | ||
[Coke] | m: my $was_in_lazy; my $var := lazy { $was_in_lazy++; 42 }; use Test; ok !$was_in_lazy, 'lazy block wasn\'t yet executed (1)'; | ||
camelia | rakudo-moar debc7c: OUTPUT«Method 'lazy' not found for invocant of class 'Int' in block <unit> at <tmp> line 1» | ||
[Coke] | m: my num $alpha = Nil; # RT #121518 | 19:03 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=121518 | ||
camelia | rakudo-moar debc7c: OUTPUT«Cannot unbox a type object in block <unit> at <tmp> line 1» | ||
[Coke] | .tell nine tagged a [PRECOMP] related bug, RT #128548 | 19:10 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128548 | ||
yoleaux2 | [Coke]: I'll pass your message to nine. | ||
masak | lizmat: thank you. I ended up with this workaround for now: github.com/masak/007/commit/ab1a69...46788bd2a6 | 19:14 | |
[Coke] | LHF: RT #124788 has some skipped tests; for many of them we can remove the skip fudge line. | 19:20 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124788 | ||
[Coke] | m: POST {} # RT # 124961 | 19:25 | |
camelia | rakudo-moar f0abe8: OUTPUT«Postcondition '{}' failed in block <unit> at <tmp> line 1» | ||
masak | \o/ | ||
m: POST { Bool.roll } | 19:26 | ||
camelia | rakudo-moar f0abe8: OUTPUT«Postcondition '{ Bool.roll }' failed in block <unit> at <tmp> line 1» | ||
masak | dangit | ||
masak blows on the dice | |||
m: POST { Bool.roll } | 19:27 | ||
camelia | ( no output ) | ||
masak | \o/ | ||
[Coke] | m: my $c; 'abc' ~~ /.{ $c = $¢ }/ ; say $c.WHAT; | ||
camelia | rakudo-moar f0abe8: OUTPUT«(Any)» | ||
[Coke] | m: my $c; 'abc' ~~ /.{ say $¢.WHAT }/ ; | ||
camelia | rakudo-moar f0abe8: OUTPUT«Nil» | ||
[Coke] | that one might just need a better test (RT #124998) | 19:29 | |
synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124998 | ||
jnthn | Hm, I did the troublesome supplies changes in a branch. | 19:45 | |
The one that went into master passed all spectets | |||
arnsholt | jnthn: Could you give my better-O branch a once-over? I think it should be good, but my last couple of quick hacks have turned out to be not so quick after all, so I figure it's better safe than sorry =) | 19:52 | |
jnthn | Moar master / NQP 255d2e9d1f1 / Rakudo 948aee0f405 just gave me a clean spectest on Linux in my VM, TEST_JOBS = 12, 4 cores available to the VM | 19:55 | |
[TuxCM] | Despite the tests failing, timing: | 20:05 | |
This is Rakudo version 2016.07.1-68-g4c848b0 built on MoarVM version 2016.07-4-g236058a | |||
test 14.863 | |||
test-t 8.004 | |||
csv-parser 16.288 | |||
jnthn | With Rakudo HEAD I also see the failures, for those following along here. :) | 20:07 | |
Given I get them with just a different Rakudo version and on the same MoarVM, I think for now we can probably count the MoarVM changes out. | |||
948aee0f405cbd seems OK | 20:08 | ||
Trying the commits since then | 20:09 | ||
Broken in 4c848b0216b4 | 20:11 | ||
TimToady: Did you spectest? | |||
Building HEAD + revert of 4c848b021 now and spectesting | 20:12 | ||
I'm suspecting that if this does come out clean, it'll be that the change in that commit breaks something in the compilation of supply/whenever blocks or some of the CORE.setting code implementing supplies. | 20:13 | ||
Early indications are that revert fixes it. | 20:14 | ||
TimToady | hmm, I did, but assumed the supplies were breaking because you were fiddling with them; guess that was a bad assumption | 20:19 | |
jnthn | Ah, ok | ||
I'll have confirmation in < 1 min | |||
Clean spectest with the revert. | 20:20 | ||
TimToady: What'd you prefer, that I push the revert, or that I leave you to fix it? | |||
(I'm too tired to dig into what it might be, I'm afraid.) | 20:21 | ||
TimToady | I'll fix it, or revert it... | 20:22 | |
jnthn | OK | ||
TimToady | hopefully just something stupid | ||
jnthn | Local revert scrubbed | ||
So I don't push it by accident... :) | |||
dalek | kudo/nom: 8ede35d | TimToady++ | src/Perl6/Actions.nqp: unbreak supplies Undo wanting of method return till we figure out why it breaks supplies. (Reintroduces spurious Useless use of 'self'.) |
20:40 | |
[TuxCM] | All tests successful. | 20:48 | |
Files=28, Tests=22288, 32 wallclock secs ( 3.13 usr 0.16 sys + 89.64 cusr 1.76 csys = 94.69 CPU) | |||
Result: PASS | |||
thank you | |||
jnthn | \o/ | ||
[TuxCM] | This is Rakudo version 2016.07.1-71-g8ede35d built on MoarVM version 2016.07-4-g236058a | 20:52 | |
test 14.869 | |||
test-t 8.028 | |||
csv-parser 15.898 | |||
japhb | Hanging within ~ half a percent of 8.0 for test-t, it seems | 20:53 | |
For the last few days, I mean | |||
jnthn | oh man, I saw the word hanging and thought "oh crap, another deadlock to find" ;P | 20:57 | |
japhb | Heh | 22:34 |