Zoffix Finally have some useful data. Comparing 100-top-modules install on 2017.04.3 rakudo and nom HEAD, same amount fails (18 on head, 17 + 1 hang on 04): gist.github.com/zoffixznet/6e8277b...f1a540f735 00:14
BUT, a lot of *different* modules fail :/
# Failed test 'Add to index one page'
# at t/storage/index.t line 28
# expected: '["Foo"]'
# got: '$["Foo"]'
failure in November for example :/
Actually 1-fewer failed on HEAD, as I have a fix for Digest 00:15
Basically 4 new module failure: Bailador GraphQL Perl6::Maven WebSocket 00:16
Might be just my testing setup. Gonna try to reinstall and stuff
ugexe GraphQL is because of Bailador 00:21
i assume
Zoffix oh and Perl6::Maven too 00:25
oh and WebSocket depends on Digest, so should be good too after I push my fix 00:26
So just Bailador basically :)
Ohh.. wait
Yes, Bailador deps on Digest \o/ so no regressing modules \o/
ugexe now start harassing p6c testers and cpan testers with the reports >:) 00:31
Zoffix :) 00:32
Geth rakudo/nom: 2f22b701d4 | (Zoffix Znet)++ | src/core/Int.pm
Add kludge to +>, <+ ops to prevent inlining bug

Works around RT#131306: rt.perl.org/Ticket/Display.html?id=131306
Fixes github.com/grondilu/libdigest-perl6/issues/18		 00:34
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131306
Geth roast: c1d62112f3 | (Zoffix Znet)++ | S03-operators/bit.t
Cover spesh bug with bit shift ops

RT#131306: rt.perl.org/Ticket/Display.html?id=131306		 00:35
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131306
Zoffix .tell jnthn for that shift-op-inlining bug ( RT#131306 ), instead of reverting previous fixes and re-introducing old bugs, I managed to work around it with the nqp::ctx thing in github.com/rakudo/rakudo/commit/2f22b701d4 00:41
yoleaux Zoffix: I'll pass your message to jnthn.
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131306
Zoffix Windows spectest failures after my fix to make-temp-file: gist.github.com/zoffixznet/9b12853...ecbe379469 00:55
ZOFVM: Files=1242, Tests=135586, 113 wallclock secs (21.65 usr 3.13 sys + 2328.31 cusr 165.67 csys = 2518.76 CPU) 01:27
Geth rakudo/nom: 4ae7c6976a | (Zoffix Znet)++ | src/core/Process.pm
Fix unwanted warnings on Windows with $*HOME

When HOME and (HOMEDRIVE or HOMEPATH) are unset		 01:29
Zoffix Well, that fixes one failure...
...dozens more to go :(
neither CTRL+Z nor CTRL+D exits REPL and it also appears to be one of the failures in t\spec\S16-filehandles\argfiles.t because closing Proc.in crashes the child process 01:49
^ I mean on Windows
Zoffix calls it a day 01:50
Geth roast: 6bad631d37 | (Zoffix Znet)++ | packages/Test/Util.pm
Re-try deleting working files in is_run

It appears on Windows, when the `is_run` code launches it's own Proc that does STDIN things, there's some delay before `.in` file can be deleted by `is_run` so if we detect failurage to delete a working file, before dying, sleep a bit and retry and only if we fail then, then die.		 01:56
AlexDaniel bisectable6ā€¦ why? 02:25
ugexe m: DEPRECATED("foo"); 03:42
camelia Indexing requires an instance, tried to do: $bt[ (Any) ]
in block <unit> at <tmp> line 1
ugexe self documenting at least 03:43
[Tux] This is Rakudo version 2017.04.3-295-g4ae7c6976 built on MoarVM version 2017.04-68-g5f233249 07:19
csv-ip5xs 2.545
test 12.764
test-t 4.225 - 4.246
csv-parser 13.169
lizmat Files=1192, Tests=58610, 205 wallclock secs (12.36 usr 4.74 sys + 1218.35 cusr 114.14 csys = 1349.59 CPU) 07:20
m: dd <a b b c c c>.Bag.pick(-3) # more strangeness :-( 07:52
camelia ("c", "c", "b", "c", "a", "b").Seq
lizmat m: dd <a b b c c c>.Bag.roll(-3) # and infiniloops
camelia (timeout)
Geth rakudo/nom: bb64d85f0b | (Elizabeth Mattijsen)++ | src/core/Baggy.pm
Baggy!ROLLPICKGRABN now returns Iterator

For easier integration with other Iterator returning logic		 08:00
rakudo/nom: 40e4e13200 | (Elizabeth Mattijsen)++ | src/core/Baggy.pm
Fix handling of <1 and NaN in Baggy.pick
 08:21
rakudo/nom: e5d2c6f67a | (Elizabeth Mattijsen)++ | src/core/Baggy.pm
Fix handling of <1 and NaN in Baggy.roll
 08:39
rakudo/nom: 44c89ed959 | (Elizabeth Mattijsen)++ | src/core/Baggy.pm
Fix handling of Baggy.pick(N) on empty Baggys
 09:02
rakudo/nom: 4b1b349424 | (Elizabeth Mattijsen)++ | src/core/Rakudo/QuantHash.pm
Introducing R:Q.TODO

Sanity check on number of items to fetch, throws on NaN		 09:52
rakudo/nom: 22ba231738 | (Elizabeth Mattijsen)++ | src/core/Baggy.pm
Fix handling of Baggy.pick(NaN) on empty Baggys

Before, emptiness was checked before NaNness
rakudo/nom: 23b62eb89c | (Elizabeth Mattijsen)++ | src/core/BagHash.pm
Simplify $count test on BagHash.grab(N)

Now that we have R:Q.TODO		 10:25
rakudo/nom: abc0509c5c | (Elizabeth Mattijsen)++ | src/core/Rakudo/QuantHash.pm
Fix various issues with Baggy.(pick|grab)pairs

  - handle empty Baggy handling (a Mu was missing)
  - handle NaN consistently with the new R:Q.TODO
rakudo/nom: 316399cf7b | (Elizabeth Mattijsen)++ | 2 files
Remove references to pand :-(
 10:44
lizmat panda rather :-(
timotimo that's pandaring 10:58
Zoffix NeuralAnomaly: status
NeuralAnomaly Zoffix, [āœ˜] Next release is today. Since last release, there are 39 new still-open tickets (2 unreviewed and 0 blockers) and 14 unreviewed commits. See perl6.fail/release/stats for details
lizmat Zoffix: about to fix the SetHash.values c.s. issues 10:59
Zoffix .tell AlexDaniel BTW, latest and greatest Rakudo fixed a bug that was affecting Digest, so you can use it now for your sha256 thing 11:01
yoleaux Zoffix: I'll pass your message to AlexDaniel.
Zoffix Ah, zostay's right.. hyper does stange things 11:03
jnthn
.oO( No, *this* is pandaring... bit.ly/2qByjCj )
yoleaux 00:41Z <Zoffix> jnthn: for that shift-op-inlining bug ( RT#131306 ), instead of reverting previous fixes and re-introducing old bugs, I managed to work around it with the nqp::ctx thing in github.com/rakudo/rakudo/commit/2f22b701d4
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131306
Zoffix m: my Pair @p = a => 1, b => 2; dd @pĀ».key
camelia Array[Pair].new("a", "b")
jnthn Zoffix: Nice cheat :-)
Zoffix :)
timotimo yowch 11:06
Zoffix Filed as rt.perl.org/Ticket/Display.html?id=131332
lizmat Zoffix: re github.com/rakudo/rakudo/commit/2f22b701d4,. wouldn't a simple nqp::ctx have been enough ? 11:08
without need to assign it ? 11:09
if there's a need to assign it, maybe binding would be better (save on a Scalar container creation?)
jnthn An nqp::ctx() without its result used will be optimized out pre-inlining, and so not block inlining. 11:13
(The optimizer knows that an unused nqp::ctx is a no-op)
lizmat jnthn: binding vs assignment ? 11:14
jnthn I don't know about assignment/binding. But we do lower lexicals that are binded to locals in some cse.
We don't do that with assignent yet
lizmat so you're saying binding may or may not block inlining
jnthn Spesh can't analyze things so well when there's a container there
lizmat ok, fair enough
jnthn But binding of a lower local it can see the bound value isn't used and rip out the whole chain, potentially
(Like, there's code there that can do that, and probably will.) 11:15
We'll fix it relatively soon and then the workaround can go
lizmat perhaps we need something like a "nqp::dontoptimize" ?
jnthn Or maybe a noinline trait that sets a routine flag 11:16
lizmat my worry is that some optimization may get smart enough to see $ignore isn't being used and rip it out
*before* the inlining bug is fixed
jnthn Heh
If there's time to implement something that sophisticated there's sure as heck time to fix the inlining bug ;) 11:17
Plus, we can shove a spectest in to cover it
lizmat yeah, but the order in which things happen, isn't always logical :-)
jnthn So if the workaround becomes ineffective by somebody making opts better, we'll know about it. :)
lizmat spectest++ :-) 11:18
jnthn True. I'll try and jump on this next week, anyways.
I have some tuits. :)
But either way, yes, we totally want a spectest for this
timotimo ... what if we had a nqp::preventinlining op :) 11:22
Zoffix huh 11:31
m: sub infix:<ei'oych8:;=$\x7Fzboxe$ieg%Femcd$zbz YzoickfEdo'bcgksk godof}k=knfkf9kybxk5+> { $^a + $^b }; say 42 ei'oych8:;=$\x7Fzboxe$ieg%Femcd$zbz YzoickfEdo'bcgksk godof}k=knfkf9kybxk5+ 70
camelia 5===SORRY!5=== Error while compiling <tmp>
Too many symbols provided for categorical of type infix; needs only 1
at <tmp>:1
------> 3YzoickfEdo'bcgksk godof}k=knfkf9kybxk5+>7ā5 { $^a + $^b }; say 42 ei'oych8:;=$\x7Fz
Zoffix "only 1"? I made infixes larger than that
Tho on this font I see there's spaces in it
AlexDaniel m: sub infix:<a b> { $^a + $^b }' 11:33
camelia 5===SORRY!5=== Error while compiling <tmp>
Too many symbols provided for categorical of type infix; needs only 1
at <tmp>:1
------> 3sub infix:<a b>7ā5 { $^a + $^b }'
yoleaux 11:01Z <Zoffix> AlexDaniel: BTW, latest and greatest Rakudo fixed a bug that was affecting Digest, so you can use it now for your sha256 thing
AlexDaniel \o/
thanks 11:35
Zoffix Oh, OK, now I see what "only 1" means. *symbol* != *character* 11:43
timotimo it could be rewritten as "parts" 11:44
and also tell how many it thinks there were
lizmat: would it make sense for hypers to return a seq? hypers are meant to mimic the input structure, and they are also the opposite of lazy 11:46
m: ((:1a, :2b), ((:3c, :4d), :9z))>>.key.perl.say 11:47
camelia ($("a", "b"), $($("c", "d"), "z"))
lizmat ah, then what is the problem with RT #131332 ?
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131332
Zoffix Is it possible to make it throw the typecheck just as if you'd try to assign strings on your own? 11:52
m: Array[Pair].new("a", "b")
camelia Type check failed in assignment to ; expected Pair but got Str ("a")
in block <unit> at <tmp> line 1
Zoffix "Plus, we can shove a spectest in to cover it" That's already done in github.com/perl6/roast/commit/c1d6...ed11996f17 11:54
m: dd Buf.new(1, 2, 3)Ā».Str 11:58
camelia ("1", "2", "3")
Zoffix hmmmm
timotimo lizmat: the problem is that it results in an Array[Pair] that contains Str instead of Pair
lizmat ah! 11:59
ok
Zoffix m: dd (1, 2, 3).SeqĀ».self
camelia (1, 2, 3)
timotimo oh, cool, i didn't know we have a self method 12:01
Geth rakudo/nom: e5b5d3461d | (Elizabeth Mattijsen)++ | src/core/SetHash.pm
Fix SetHash.(iterator|values|kv)

  - simplification
  - give the same semantics as (Bag|Mix)Hash.(iterator|values|kv)		 12:02
lizmat Zoffix: I think ^^^ fixed the issues you found with SetHash 12:06
I *think* I covered all of the other issues you mentioned in earlier commits today 12:07
all that is needed, are more tests
which I will start working on now
Zoffix lizmat++ thanks
lizmat m: my $s = <a b c>.SetHash; for $s.values { $_ = 0; last; $_ = 1 }; dd $s # now also lazy 12:21
camelia SetHash $s = SetHash.new("a","c")
lizmat m: my $s = <a b c>.SetHash; for $s.values { $_ = 0; $_ = 1 }; dd $s # now also recovering 12:25
camelia SetHash $s = SetHash.new("a","c","b")
dogbert17 m: say āˆ… āŠ‚ (1) 12:36
camelia True
dogbert17 m: say āˆ… āŠ‚ () # is this correct? 12:37
camelia False
lizmat I think the empty set is a subset of any set
m: say āˆ… āŠ‚ āˆ…
camelia False
lizmat hmmm....
m: say āˆ… āŠ‚ set()
camelia False
lizmat m: say āˆ… āŠ‚ set(1)
camelia True
lizmat m: say āˆ… āŠ‚ (1).set 12:38
camelia No such method 'set' for invocant of type 'Int'
in block <unit> at <tmp> line 1
lizmat m: say āˆ… āŠ‚ (1).Set
camelia True
dogbert17 commit: 2016.12 say āˆ… āŠ‚ set()
committable6 dogbert17, Ā¦2016.12: Ā«FalseĀ»
dogbert17 doesn't look new
lizmat indeed
please RT it, will look at it 12:39
dogbert17 consider it done :)
Zoffix m: say āˆ… āŠ† set() 12:42
camelia True
Zoffix m: say 0 < 0
camelia False
Zoffix m: say 0 <= 0
camelia True
dogbert17 lizmat: RT #131335 12:44
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131335
Zoffix Also: www.wolframalpha.com/input/?i=%E2%...+%E2%88%85 12:45
Geth roast: 8cd62dba84 | (Elizabeth Mattijsen)++ | S02-types/baghash.t
Add tests for BagHash.(values|kv|pairs) modifying
 12:49
dogbert17 so if āŠ‚ signifies a 'proper subset' then my report is bogus?
lizmat perhaps 12:50
en.wikipedia.org/wiki/Empty_set 12:51
dogbert17 m: say (1) āŠ‚ (1) 12:53
camelia False
dogbert17 m: say (1) āŠ† (1) 12:54
camelia True
dogbert17 seems to me that āŠ‚ should be interpreted at 'proper subset' 12:55
Zoffix m: say āˆ… āŠ‚ set āˆ… 12:56
camelia True
Geth roast: f80ad6c9b9 | (Elizabeth Mattijsen)++ | S02-types/mixhash.t
Add tests for MixHash.(values|kv|pairs) modifying
 12:57
dogbert17 m: say āˆ… āŠ‚ āˆ… 13:01
camelia False
dogbert17 m: say āˆ… āŠ‚ set āˆ… # why the difference?
camelia True
Zoffix dogbert17: one is a set() āŠ‚ set(), the other is a set() āŠ‚ set(set()) 13:02
dogbert17 aha 13:03
Zoffix NeuralAnomaly: status 13:04
NeuralAnomaly Zoffix, [āœ˜] Next release is today. Since last release, there are 43 new still-open tickets (4 unreviewed and 0 blockers) and 15 unreviewed commits. See perl6.fail/release/stats for details
dogbert17 leaning towards rejecting my own RT ...
Zoffix I guess it's time \o/
Zoffix finishes up a level of Trine first... 13:05
dogbert17 waits for lizmat's judgement
lizmat hmmm.... re set() being a subset of set() ? 13:06
masak or jnthn would be a better judge of that, I think
Zoffix: writing new SetHash tests now, found an issue :-( 13:07
ah, the issue appears to be in the test :-) 13:09
Geth roast: b580fffa36 | (Elizabeth Mattijsen)++ | 2 files
Add tests for SetHash.(values|kv|pairs) modifying
 13:10
Zoffix If we assume āˆ… āŠ‚ āˆ… => True, that means there's $x we can put into āˆ… to make āˆ…, since there is no such $x, then False is the right answer. WolframAlpha also says the False is the answer. 13:11
lizmat ah, actually, a subset always implies fewer elements on the left than on the right 13:12
since the empty set does not have fewer elements than the empty set, it is by definition not a subset 13:13
so the current behaviour appears correct
dogbert11 lizmat: the 'subset with fewer elements' is what is often called 'proper subset' 13:14
lizmat ok, and it appears that's how it is implemented
dogbert11 guru4711 13:15
:(
lizmat time to change the pw ?
dogbert11 yup
I'll close the report, sorry for the noise
lizmat no pb, we need inquisitive minds :-) 13:16
and they need to know :-)
Zoffix: thinking of stopping writing tests right now, so you can do a release ;-) 13:17
Zoffix OK 13:20
dogbert17 and I can't seem to close my own ticket (sigh) 13:21
I could :)
Zoffix takes a final look over Win32 spectest failures and starts cutting release after that 14:03
C:\release-rakudo>perl6 -e "my $proc = Proc::Async.new: $*EXECUTABLE, '-e', 'say 42'; $proc.stdout.tap: { say qq|Meow: $_| }; await $proc.start" 14:09
===SORRY!===
Argument to "say" seems to be malformed
Don't know what it's on about ĀÆ\_(惄)_/ĀÆ
Seems Test::Util's doesn't-hang is busted on Windows
C:\release-rakudo>perl6 -e "with Proc::Async.new($*EXECUTABLE, '-e', '$*OUT.write(Blob.new(65, 66, 67, 13, 10))') { .stdout.tap: *.say; await .start }" 14:13
===SORRY!=== Error while compiling -e
Unable to parse expression in argument list; couldn't find final ')'
Or seems it's Proc::Async that's busted on windows :/
which doesn't-hang usess 14:14
Was already busted in last release, so I guess I'll skip it.
yoleaux Zoffix: Zoffix: ensure these (and and few lines up/below) issues are resolved/filed before release: irclog.perlgeek.de/perl6-dev/2017-...i_14607286 14:17
Zoffix botsnack
synopsebot6 om nom nom
Zoffix Looks committable6 has a race in it. If you give it code to run too fast, it hangs 14:20
lizmat goes afk for a few hours 14:22
MasterDuke Zoffix: can you create a whateverable issue? 14:29
Zoffix github.com/perl6/whateverable/issues/148 14:42
MasterDuke Zoffix++ 14:43
Geth roast: 998f3d45b3 | (Zoffix Znet)++ | S05-match/capturing-contexts.t
Add more tests for <( and )>
 14:54
Zoffix m: my $b = <a b b c c c>.SetHash; $_ = -1 for $b.values; dd $b 14:56
camelia SetHash $b = SetHash.new("b","a","c")
Zoffix m: my $b = <a b b c c c>.BagHash; $_ = -1 for $b.values; dd $b
camelia BagHash $b = ().BagHash
Zoffix lizmat: are you around by any chance? Was ^ SetHash version meant to also return ().SetHash above? 14:57
Like this line github.com/rakudo/rakudo/blob/e5b5...ash.pm#L35 14:59
Should it be nqp::isgt_i($value,0) ?
c: 2017.04.3 my $b = <a b b c c c>.BagHash; $_ = -1 for $b.values; dd $b
committable6 Zoffix, Ā¦2017.04.3: Ā«BagHash $b = ("b"=>-1,"a"=>-1,"c"=>-1).BagHashĀ»
Zoffix c: 2017.04.3 my $b = <a b b c c c>.SetHash; $_ = -1 for $b.values; dd $b 15:00
committable6 Zoffix, Ā¦2017.04.3: Ā«SetHash $b = SetHash.new("b","a","c")Ā»
Zoffix Guess we can figure it out after release...
"give the same semantics as (Bag|Mix)Hash.(iterator|values|kv)" e5b5d3461d300fe4ae36d3a 15:01
So then yeah, it's a typo. /me fixes
Geth rakudo/nom: e9ae004751 | (Zoffix Znet)++ | src/core/SetHash.pm
Fix SetHash internal `proxy` sub from treating negatives as positive

Adjusts behaviour to follow commit message in previous[^1] commit about giving SetHash same semantics as Bag/MixHashes; such as:
  `$_ = -1 for $sethash.values` deleting elements instead of leaving them in.
  [1] github.com/rakudo/rakudo/commit/e5...13c222ab26		 15:11
roast: cf06354b88 | (Zoffix Znet)++ | S02-types/sethash.t
Test negatives assigned to .value delete from SetHash

Rakudo fix: github.com/rakudo/rakudo/commit/e9ae004751		 15:15
Zoffix crap 15:22
6.c-errata fails in t/spec/integration/advent2011-day07.t with some weird error: gist.github.com/zoffixznet/841902c...b8f09e6333 15:23
Zoffix-- # forgetting to test 6.c-errata in the past few days :(
dogbert17 oops 15:24
Geth rakudo/nom: 35df8adc1e | (Zoffix Znet)++ | docs/ChangeLog
Log all commits to date
 15:26
Zoffix weird.. if I add `$ = $meth.^name;` after this line github.com/perl6/roast/blob/master...iler.pm#L8 then the test passes :S 15:29
MasterDuke i think there was some conversation about that error recently
Zoffix oh wait
it passes on my local box but fails on my google VM
MasterDuke: where?
MasterDuke here
Zoffix Then when? I don't recall it and I read pretty much everything 15:30
MasterDuke couple weeks ago maybe?
irclog.perlgeek.de/perl6-dev/2017-...i_14525903 15:31
Zoffix Ah ok. I see Apr fixes in master 15:32
I thought April was ancient history, but yeah, it was after last release..
*phew* :)
*insert obligatory grumble about people mending tests in master but forgetting to do the same in errata* 15:33
Geth roast/6.c-errata: 1260b6c88a | TimToady++ (committed by Zoffix Znet) | 2 files
fix for correct BUILD semantics of grammar

This test uses a module that lists a bunch of internal methods. This is suboptimal, but for now just add the methods that are called when a proper new/BUILDALL are done. (The old code only worked because class Grammar bypassed all that and went straight to nqp, which in turn made it impossible to support user grammars with their own attributes.) ... (5 more lines)
roast/6.c-errata: 4aa4064d27 | (Elizabeth Mattijsen)++ (committed by Zoffix Znet) | 2 files
Fix final tests broken by fix to RT #131241
 15:36
roast/6.c-errata: 9dc81f0f69 | (Elizabeth Mattijsen)++ (committed by Zoffix Znet) | 2 files
Fix more tests broken by fix to RT #131241
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131241
roast/6.c-errata: a45360edbb | (Elizabeth Mattijsen)++ (committed by Zoffix Znet) | 2 files
Fix tests broken by fix to RT #131241
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131241
Zoffix k. it's time 15:39
ugexe last call on o/u for hotfixes
Zoffix "o/u"? 15:40
ugexe over/under.. its a gambling term basically saying you want this or that side of some number
MasterDuke Zoffix: forgot to mention this earlier, might suggest an edit to the changelog
"Added primary and foreign key constraints to profiler's SQL output [c776c087]". also some of the field names were changed 15:41
Geth rakudo/nom: 8a763d0be9 | (Zoffix Znet)++ | docs/ChangeLog
Change ChangeLog Entry

Per irclog.perlgeek.de/perl6-dev/2017-...i_14611505		 15:43
MasterDuke Zoffix++
Zoffix NeuralAnomaly: status 15:44
NeuralAnomaly Zoffix, [āœ˜] Next release is today. Since last release, there are 41 new still-open tickets (0 unreviewed and 0 blockers) and 3 unreviewed commits. See perl6.fail/release/stats for details
Zoffix NeuralAnomaly: status
NeuralAnomaly Zoffix, [āœ”] Next release is today. Since last release, there are 41 new still-open tickets (0 unreviewed and 0 blockers) and 0 unreviewed commits. See perl6.fail/release/stats for details
Zoffix NeuralAnomaly: cut the release
hmm
NeuralAnomaly: cut the release 15:45
NeuralAnomaly Zoffix, Will do! If you're feeling particularly naughty, you can watch me at perl6.fail/release/progress or go look at some cats icanhas.cheezburger.com/lolcats
Zoffix, ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜  ABNORMAL EXIT!
Zoffix NeuralAnomaly: cut the release 15:46
NeuralAnomaly Zoffix, Will do! If you're feeling particularly naughty, you can watch me at perl6.fail/release/progress or go look at some cats icanhas.cheezburger.com/lolcats
Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ Prep done
Geth nqp: 982acbda15 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
bump MoarVM version to 2017.05
 15:47
nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2......2017.05
0d2e99a8dc | (Zoffix Znet)++ | VERSION
Ā¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2......2017.05
Zoffix tbrowder: RE github.com/perl6/doc/commit/9e5d047f27 isn't `experiemental` user-facing? like to include experimental features. Never heard of newline, but it might be too 15:53
NeuralAnomaly Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ nqp tests OK 15:57
tbrowder hm, maybe we should eliminate the two categories and reword the lead-in a bit? 15:58
Zoffix maybe
tbrowder i'll give it a shot.... 15:59
MasterDuke is there a way to run the nqp tests in parallel?
NeuralAnomaly Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ nqp release tarball tests OK 16:08
Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ nqp release DONE
MasterDuke hm, jnthn said something about `--jobs=20` here irclog.perlgeek.de/moarvm/2017-05-..._14542434, but that doesn't have any effect for me
Geth rakudo/nom: 38441cb5a6 | (Zoffix Znet)++ | docs/announce/2017.05.md
Generate release announcement for 2017.05
rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017.......2017.05
756645b3bf | (Zoffix Znet)++ | tools/build/NQP_REVISION
rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017.......2017.05
f609e69e1a | (Zoffix Znet)++ | VERSION
Ā¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017.......2017.05
Zoffix MasterDuke: I see --jobs arg in `prove`, so I would've thought that'd do something 16:10
MasterDuke `make --jobs=10 m-test` takes exactly the same amount of time as without the `--jobs=10` 16:12
ugexe does that send --jobs=10 to prove? 16:13
NeuralAnomaly Zoffix, ā˜ ā˜ ā˜ ā˜ ā˜ ā˜  Rakudo: install Inline::Perl5 16:14
Zoffix, ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜  ABNORMAL EXIT!
Zoffix boo
"fetch stage failed for Inline::Perl5: Unable to handle source 'www.cpan.org/authors/id/N/NI/NINE/P...6.tar.gz'"
wtf :/
and by wtf I mean "I've just did a clean install on the same box an hour ago and it was fine"
Ohhh 16:15
NeuralAnomaly's script is still using panda
Stupid robot. Don't you follow the news? 16:16
mst LOL
MasterDuke ugexe: looks like it does not. if i run prove manually with `--jobs=10` it takes one-fifth the time
mst well, no, --jobs is for *make* jobs
ugexe just rewrite bin/panda as a zef front end
mst that's obviously not going to work
MasterDuke mst: but it does
mst MasterDuke: lol? 16:17
MasterDuke: no it doesn't, you already said it didn't
again: 'make --jobs=10' is obviously not going to work
MasterDuke using make it doesn't do anything, running prove manually it does
mst yes, that's what I'm telling you
because make also accepts a --jobs argument for the number of *make* jobs to run 16:18
prove is run by a *single* make job
mst wonders if you're even reading what I'm writing here :(
MasterDuke yeah. any way to get that option passed to prove?
ugexe isn
isnt there some -j10 variant for env?
to just do TAP_HARNESS_J=10 make whatever
MasterDuke mst: i thought i was understanding what you were saying 16:19
mst I was explaining why 'make --jobs=10' was irrelevnt because that's an argument to make, which does something different to the argument to prove
also, where's the Makefile?
MasterDuke TEST_JOBS=n for rakudo, but that doesn't do anything for NQP 16:20
mst try HARNESS_OPTIONS=j9
MasterDuke mst: it's built by Configure.pl, there isn't one in the NQP repo
Zoffix NeuralAnomaly: steps
NeuralAnomaly Zoffix, all pre nqp r post pre-r6 pre-blank-slate nqp-clone nqp-bump-vers nqp-build nqp-test nqp-tar nqp-tar-build nqp-tag nqp-tar-sign nqp-tar-copy r-clone r-prep-ann r-bump-vers r-build r-p5 r-stress r-stress-v6c r-tar r-tar-build r-tar-p5 r-tar-stress r-tag r-tar-sign r-tar-copy post-scp
Zoffix NeuralAnomaly: run r-p5 r-stress r-stress-v6c r-tar r-tar-build r-tar-p5 r-tar-stress r-tag r-tar-sign r-tar-copy post-scp 16:21
MasterDuke mst++ that worked
mst there might equally be a way to get the Makefile to pass it to prove ... but I don't think we care now 16:23
though you might want to add a note in the docs somewhere
MasterDuke heh, was just going to ask that
NeuralAnomaly Zoffix, ā˜ ā˜ ā˜ ā˜ ā˜ ā˜  Rakudo: make stresstest (master) 16:24
Zoffix, ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜  ABNORMAL EXIT!
Zoffix NeuralAnomaly: yo 16:28
NeuralAnomaly Zoffix, Yo, yo, yo! What up, homeslice?
mst MasterDuke: well, I could look, but you still didn't show me the makefile like I asked, so I can't
Zoffix What's up with that error...
ugexe so exit. much abnormal 16:29
Zoffix wishes the bot had a command to execut arbitrary shell command :/ 16:30
mst yes because that would be a totally safe and sane idea
MasterDuke mst: my NQP makefile gist.github.com/MasterDuke17/e92c9...20430e8f83 16:31
Zoffix mst: it would :) 'cause the IP for the box the bot uses is valid for a short time period so even if someone managed to run a command as admin, the box would be offline 16:34
mst MasterDuke: make PROVE='prove -j9' looks like it'd work? 16:35
(not sure if that's really an improvement but still)
Zoffix NeuralAnomaly: steps 16:36
NeuralAnomaly Zoffix, all pre nqp r post pre-r6 pre-blank-slate nqp-clone nqp-bump-vers nqp-build nqp-test nqp-tar nqp-tar-build nqp-tag nqp-tar-sign nqp-tar-copy r-custom r-clone r-prep-ann r-bump-vers r-build r-p5 r-stress r-stress-v6c r-tar r-tar-build r-tar-p5 r-tar-stress r-tag r-tar-sign r-tar-copy post-scp
Zoffix NeuralAnomaly: run r-custom 16:37
NeuralAnomaly Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ Rakudo: custom command
MasterDuke mst: yep, that works also, thanks
NeuralAnomaly Zoffix, ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜  ABNORMAL EXIT!
Geth roast: afead9ccad | (Zoffix Znet)++ | S32-io/io-handle.t
Fix tty detection in new test
 16:39
MasterDuke any reason we couldn't/shouldn't add TEST_JOBS support to NQP? pass it's value to prove's -j?
Zoffix MasterDuke: that seems the best option. 16:40
I already have that env var set due to its use in rakudo
NeuralAnomaly: run r-stress r-stress-v6c r-tar r-tar-build r-tar-p5 r-tar-stress r-tag r-tar-sign r-tar-copy post-scp 16:41
MasterDuke i can give it a try, but my make-fu is weak and my last Configure.pl/Makefile commit got reverted 16:42
Zoffix try and try again :)
NeuralAnomaly Zoffix, ā˜ ā˜ ā˜ ā˜ ā˜ ā˜  Rakudo: make stresstest (master) 16:43
Zoffix, ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜ ā˜  ABNORMAL EXIT!
Zoffix *sigh*
ZOFFLOP: t/spec/S17-supply/supplier-preserving.t
NeuralAnomaly: run r-stress r-stress-v6c r-tar r-tar-build r-tar-p5 r-tar-stress r-tag r-tar-sign r-tar-copy post-scp
c'mon, robot! I ain't got all day
ZofBot: give a hand to your pal, don't just stand there! 16:45
ZofBot Zoffix, IO; $p
NeuralAnomaly Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ Rakudo stresstest (master) OK 16:46
Zoffix yey
NeuralAnomaly Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ Rakudo stresstest (6.c-errata) OK 16:48
Zoffix damn door-to-door spammers... Why yes, I'd love to talk to a random dude with an iPad about improving my Internet connection speeds while wearing nothing but underwear 16:54
NeuralAnomaly Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ Rakudo release DONE 16:56
Zoffix, ????????????????????????????????????????????????????????????
Zoffix, ā™„ā™„ā™„ā™„ā™„ā™„ Post: upload tarballs to rakudo.org and rakudo.perl6.org
Zoffix, The release of **Rakudo #111 2017.05** has now been completed
Zoffix, ????????????????????????????????????????????????????????????
NeuralAnomaly celebrates with an appropriate amount of fun
Zoffix weeee
NeuralAnomaly: thanks
NeuralAnomaly Zoffix, any time, buddy!
mst your bots are fucking ridiculous 16:57
trout.me.uk/mstcat.jpg
Zoffix :) 16:58
AlexDaniel
.oO( bye-bye bot )
 16:59
Zoffix Just placed into "safe" mode... just in case... :) github.com/zoffixznet/na/commit/4e...3a64c1241a 17:00
AlexDaniel slaps committable6 17:01
committable6: 96 open issues and you solved none
committable6 AlexDaniel, Ā¦96: Ā«Cannot find this revision (did you mean ā€œv6.cā€?)Ā»
Geth rakudo/nom: 598361a943 | (Zoffix Znet)++ (committed using GitHub Web editor) | docs/release_guide.pod
2017.05 is now in the past
 17:04
Zoffix NeuralAnomaly: status 17:05
NeuralAnomaly Zoffix, [āœ˜] Next release will be in 3 weeks and 6 days. Since last release, there are 3 new still-open tickets (0 unreviewed and 0 blockers) and 4 unreviewed commits. See perl6.fail/release/stats for details
mst Zoffix: ohhh, right, it only runs stuff at all when in a "release" mode?
Zoffix mst: yeah + I need to tell it the IP of my google VP it connects to, and that IP changes all the time
s/VP/VM/; 17:06
mst ok, then I see why you suspect it's probably safe enough
Zoffix :)
MasterDuke is ?= supported in makes other than GNU Make? 17:08
mst, geekosaur: ^^^ ? 17:11
geekosaur not generally
MasterDuke sonofa 17:13
geekosaur now you see why so many projects just insist you use gnu make
mst Get New Utilities 17:15
MasterDuke how about `$(or $X,$X,abc)`? 17:17
geekosaur all make "functions" are gnu extensions 17:18
bsdmake has completely different functions; nmake doesn't have any that I know of
MasterDuke ok, so the problem i have is that if i just do `PROVE = 'prove -j$(TEST_JOBS)'`, if you don't specify TEST_JOBS then prove complains 17:20
what's a cross-make way to pull a value from an env variable, but give a default if that env variable isn't defined?
geekosaur if I were targeting unix only I'd use the shell version. but I don't think that flies with nmake either 17:21
(it doesn't; naturally it uses CMD.EXE syntax) 17:23
you're going to have to do this by forcing it to be defined before running it, in whatever platform appropriate way 17:25
MasterDuke don't quite follow
geekosaur: does (n|bsd)make support this: `$(shell expr $(TEST_JOBS) + 0)`? 17:37
lizmat m: say "truthy" if -1 # Zoffix, re SetHash.values $_ = -1 17:42
camelia truthy
lizmat so yes, it is supposed to be that way
afk again&
geekosaur neither does $(shell) which is a make function which is gnumake-only as I already said 17:43
and as I also already said, nmake uses CMD.EXE not a Unix shell 17:44
MasterDuke ugh. i may just leave this alone and make sure to have the non-TEST_JOBS NQP make command in my Ctrl-r history 17:45
i don't have (n|bsd)make to test with, nor do i know enough to write something that i'm certain will work with them 17:47
geekosaur unfortunately you can't use 0 as the default for that, so you don't get to use about the only hack that would be portable: -j0$(TEST_JOBS) 17:48
MasterDuke OH! that does actually work 18:02
geekosaur: why did you think that wouldn't work? it does for me: `make m-test` runs `prove -j0` and that acts just like `prove`, and `TEST_JOBS=10 make m-test` runs `prove -j010` which does do multiple jobs in parallel 18:06
geekosaur because -j0 often means "as many as possible" instead of meaning the same as -j1. if that is what you want, fine. if not, you have a problem 18:07
(want to be the default, that is)
MasterDuke well, does prove work the same across platforms? because for me `prove -j0` runs just one job at a time 18:09
geekosaur how many cpus does the os think you have? 18:10
MasterDuke 4 on my ubuntu machine, 8 on my arch machine. same behavior on both
geekosaur ok. I don;t know enough about prove to know what it will do but I recall someone (zoffix I think) running into this earlier and finding corner cases 18:11
the number of cpus matters because, for example, make somewhat infamously treats -j0 or bare -j as use as many cpus as possible... but some (incuding different gmake versions!) take that as # cpus, some as #cpus - 1, some as "just spawn as many jobs as possible and let the OS scheduler sort it out" 18:12
(granted, not many of the latter, but at least one gmake version does do that... and apple shipped it on several OS X versions) 18:13
oh, and # cpus + 1 (on the theory that at least one will be blocked on I/O and available to run something CPU-bound) 18:14
MasterDuke Zoffix: have you ever ran into any problems running `prove -j0`? 18:15
Zoffix never heard of that option even
yoleaux 17:22Z <araraloren> Zoffix: Em, thanks, I'll look at `upload to CPAN`
MasterDuke quotable6: /prove .+ '-j0'/ 18:17
quotable6 MasterDuke, Found nothing! 18:18
MasterDuke quotable6: /prove .+ '--jobs=0'/
quotable6 MasterDuke, Found nothing! 18:19
MasterDuke well, since prove is written in Perl i'm going to assume it'll work the same cross-platform 18:21
i'll go ahead and commit since it's a simple change, we can revert if necessary 18:22
Geth nqp: 728ca975d1 | MasterDuke17++ | tools/build/Makefile-common.in
Add TEST_JOBS=n support to NQP's `make test`
 18:26
MasterDuke mst++ geekosaur++
mst MasterDuke: please do it with a PROVE_OPTIONS variable otherwise now somebody overriding which prove binary to use will be fucked 18:27
Zoffix ZOFFLOP: t/spec/S17-supply/supplier-preserving.t 18:29
Geth rakudo/nom: a36f0c4487 | (Zoffix Znet)++ | src/core/SetHash.pm
Revert "Fix SetHash internal `proxy` sub from treating negatives as positive"

This reverts commit e9ae004751705bd4220ad951173249b0bc9cc1a1.
Per: irclog.perlgeek.de/perl6-dev/2017-...i_14611927
roast: 360378c3a4 | (Zoffix Znet)++ | S02-types/sethash.t
Change test to test opposite behaviour

The test was added earlier today and isn't part of 6.c language. Test per: irclog.perlgeek.de/perl6-dev/2017-...i_14611927		 18:30
mst comments on the commit 18:31
Geth nqp: af678371e2 | MasterDuke17++ | tools/build/Makefile-common.in
Add PROVE_OPTIONS variable

So you can override the prove binary but still get TEST_JOBS support.		 18:38
MasterDuke nqp: say(nqp::iseq_i(2**62, 2**62-1)) 19:26
camelia 1
MasterDuke m: use nqp; say(nqp::iseq_i(2**62, 2**62-1)) 19:27
camelia 0
Zoffix m: dd [ 2**62, 2**62-1 ]
camelia [4611686018427387904, 4611686018427387903]
Zoffix nqp: say(2**62); say(2**62-1) 19:28
camelia 4.61168601842739e+18
4.61168601842739e+18
MasterDuke nqp: my int $a := 2**62; my int $b := 2**62-1; say($a); say($b); say(nqp::iseq_i($a, $b)) 19:29
camelia 4611686018427387904
4611686018427387904
1
MasterDuke nqp: my int64 $a := 2**62; my int64 $b := 2**62-1; say($a); say($b); say(nqp::iseq_i($a, $b)) 19:30
camelia 4611686018427387904
4611686018427387904
1
MasterDuke timotimo, jnthn: ^^^ i presume that's not expected? 19:31
jnthn It's probably 'cus NQP does math as floating point in a bunch of cases, including apparently this one where it makes no sense at all 19:48
MasterDuke heh. i was trying out the even-moar-jit branch and had added eq_i to it, which broke nqp and rakudo, and thought i had an easy test case, but decided to double check what normal moarvm+nqp did 19:52
nqp: say(nqp::iseq_i(4611686018427387904, 4611686018427387903)) 19:55
camelia 1
MasterDuke huh, i assumed it was something during the parsing and conversion of the exponential form 19:56
decint and string_to_int both have the correct values (no surprise there), but --target=ast gives `...QAST::Op(iseq_i) - QAST::IVal(4611686018427387904) - QAST::IVal(4611686018427387904)...` 20:12
Geth rakudo/nom: 5528b10644 | (Elizabeth Mattijsen)++ | src/core/Rakudo/QuantHash.pm
Shortcut random value determination

Natives automatically coerce from / to num / int		 20:55
rakudo/nom: d28540bea2 | (Elizabeth Mattijsen)++ | src/core/Setty.pm
Make SetHash.grab about 1.8x faster

By using R:Q.ROLL, which saves us from having to create a whole list of keys to randomly select one from. Based on grabbing 26x from a 26 elem SetHash.
lizmat which concludes my hacking for today 20:56
good night, #perl6-dev!
almost forgot: Zoffix++ for the 2017.05 release!
MasterDuke nqp: my int $a := 100000000000000009; say($a) 21:41
camelia 100000000000000016
MasterDuke weird. inside IVal.new it's correct, but is wrong when IVal.value is called 21:42
geekosaur there's so many places that could happen... 21:47
MasterDuke it's the bindattr_i 21:48
right after this line in IVal.new: `nqp::bindattr_i($node, QAST::IVal, '$!value', $value);` 21:49
i added: `note("IVal.new: " ~ $value); note("IVal.new: " ~ nqp::getattr_i($node, QAST::IVal, '$!value'));` 21:50
the note of $value has the right number. the note of the getattr_i has the wrong number
wrong value in P6opaque.bind_attribute 22:17
timotimo, jnthn: right before this line ( github.com/perl6/nqp/blob/master/s...Val.nqp#L7 ) i put `note("IVal.new: " ~ $value);` and right after it i put `note("IVal.new: " ~ nqp::getattr_i($node, QAST::IVal, '$!value'));` 22:52
and in this if ( github.com/MoarVM/MoarVM/blob/mast...#L367-L370 ) i put `fprintf(stderr, "bi: %li\n", value_reg.i64);` 22:54
and then ran `./nqp-m -e 'my int $a := 100000000000000009; say($a)'`
the first note prints 100000000000000009, the print in bind_attribute and the second note both print 100000000000000016 22:55
if anybody has an idea of how to debug further i'm all ears 22:58

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!