TimToady m: printf "%3d", 42 00:10
camelia rakudo-moar 7f245f: OUTPUT« 42»
TimToady m: printf "%3.0d", 42
camelia rakudo-moar 7f245f: OUTPUT« »
TimToady ^^^ bug
is supposed to suppress output only if the value is 0 00:11
brokenchicken clicks "Take" 00:44
m: dd sprintf "%.3d", 55 00:53
camelia rakudo-moar 7f245f: OUTPUT«"55"␤»
brokenchicken TimToady: that's broken too, innit ^? It's supposed to pad with zeros? 00:54
and give "055"
MasterDuke m: dd sprintf "%03d", 55 00:56
camelia rakudo-moar 7f245f: OUTPUT«"055"␤»
TimToady you need explicit 0 to pad there, I think 00:58
brokenchicken m: say "|{sprintf "%10.5d", 42}
camelia rakudo-moar 7f245f: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Unable to parse expression in double quotes; couldn't find final '"' ␤at <tmp>:1␤------> 3say "|{sprintf "%10.5d", 42}7⏏5<EOL>␤»
brokenchicken m: say "|{sprintf "%10.5d", 42}"
camelia rakudo-moar 7f245f: OUTPUT«| 42␤»
brokenchicken $ perl -wlE 'say "|" . sprintf("%10.5d", 42) . "|"'
| 00042|
$ perl -wlE 'say "|" . sprintf("%.5d", 42) . "|"' 00:59
|00042|
Are we following what Perl 5 does?
TimToady huh 01:00
brokenchicken from perldoc: "For integer conversions, specifying a precision implies that the output of the number itself should be zero-padded to this width, where the 0 flag is ignored:"
geekosaur that's also ANSI C
replacing K&R's 0 flag 01:01
brokenchicken m: dd sprintf "%.3d", 123456789 01:02
camelia rakudo-moar 7f245f: OUTPUT«"123"␤»
brokenchicken heh :)
TimToady that's also wrong 01:03
should be minimum for d
brokenchicken will fix 01:04
TimToady it can be inferred from the system sprintf docs that a %.3d should have minimum 3 digits, so obviously those extra digits would have to be 0
Geth p: 92bf17d5f8 | (Zoffix Znet)++ | t/hll/06-sprintf.t
Remove outdated comment

The file is already in t/hll		 01:58
p: 1de45e4a3d | (Zoffix Znet)++ | 2 files
Fix sprintf precision parameter for `d`

Currently, it just chops the number to $precision digits, producing wrong results. The correct behaviour is for the precision to signify how many digits long the number should be, prefixing it with zeros if needed. This also results in an empty string if precision is zero and the number is zero too.
Bug find: irclog.perlgeek.de/perl6-dev/2017-...i_13966753		 02:08
MasterDuke m: dd sprintf "%0۳d", 55 02:18
camelia rakudo-moar 7f245f: OUTPUT«Directive d not applicable for type Int␤␤»
Geth p: ed1b835a7a | (Zoffix Znet)++ | t/hll/06-sprintf.t
Add tests to sprintf "%d" precision when given as "*"
 02:22
MasterDuke m: dd sprintf "%03d", 55
camelia rakudo-moar 7f245f: OUTPUT«"055"␤»
MasterDuke TimToady, brokenchicken: ^^^ should Nd's be allowed in sprintf format strings? 02:23
brokenchicken shrugs 02:24
Geth kudo/nom: a1c7d01502 | (Zoffix Znet)++ | tools/build/NQP_REVISION
(nqp bump) Fix precision for "%d" in sprintf

Fixed in github.com/perl6/nqp/commit/1de45e...8a9ce2bbe0		 02:34
rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....0-ged1b835
ast: 4e3fd00b26 | (Zoffix Znet)++ | S32-str/sprintf.t
Test sprintf "%d" precision parameter

Rakudo fix: github.com/rakudo/rakudo/commit/a1c7d01502 NQP fix: github.com/perl6/nqp/commit/1de45e4a3dcd8a9ce		 02:38
[Tux] This is Rakudo version 2017.01-7-ga1c7d0150 built on MoarVM version 2017.01-10-g76872dbe 10:09
csv-ip5xs 2.960
test 12.644
test-t 5.062
csv-parser 13.419
Geth kudo: usev6++ created pull request #1002:
Don't mention *CORE.setting in error message		 11:52
kudo/nom: 10bcec2865 | usev6++ | src/core/Backtrace.pm
Don't mention *CORE.setting in error message

References to CORE.setting.moarvm where removed with 4ae08e0b24, but rakudo-j points to gen/jvm/CORE.setting (no .jar there).
Fixes RT #130509 and RT #124767 on JVM		 11:53
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130509
Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124767
kudo/nom: a56a895808 | lizmat++ | src/core/Backtrace.pm
Merge pull request #1002 from usev6/rt130509_jvm

Don't mention *CORE.setting in error message
kudo/nom: c3c86dcf51 | (Elizabeth Mattijsen)++ | src/core/Backtrace.pm
Emphasize this is JVM backend specific behaviour
 12:17
bartolin thanks :-) 12:21
lizmat bartolin++ # for chasing JVM issues! 12:30
dogbert17 lizmat: is gist.github.com/dogbert17/b052a9da...990cdad35f affected by the grep deopt made recently? Runs twice as slowly on head compared to 2016.12 and older 12:41
lizmat dogbert17: yeah, I saw it
but if we want to support redo in a grep, there is no way around it that I can see at the moment
dogbert17 just wanted to make sure it wasn't something else, i.e. something unintended 12:42
lizmat it sucks, since redo is dynamic, so we can't even inspect the block in the optimizer for use of redo
the only way that jnthn pointed out, was to have goto working :-) 12:43
then we wouldn't need to penalize all greps for the use of redo
dogbert17 so it might be fixed in time then?
lizmat yes, once we have goto 12:44
dogbert17 cool
lizmat fwiw, this would benefit *all* map processing, not just grep
because everywhere we need to support redo, we need to build another layer of loop around everything to support it
$redo = 1; while $redo { $redo = 0; call block } 12:45
dogbert17 will it go back to the speed it had in 2016.12 ot will it be even faster with goto
lizmat it should go back to the speed we had, not faster
timotimo did you see how i built redo for whenever blocks? 12:47
lizmat timotimo: not yet
timotimo i put the loop in separately, only to be entered once we actually hit the redo handler once
perhaps that could be done here as well
i.e. { call block; CONTROL { when 'redo' { $redo = 1; while $redo { $redo = 0; call block } } } } 12:48
i *think* we only pay if redo was actually used if the code is like that
but i haven't measured
dogbert17 timotimo lizmat: should this snippet hang? 12:58
m: m: class SQLString { }; my $stringy = Str.^find_method("Stringy"); my $handler = $stringy.wrap(method () { SQLString.new(:str(callsame)) }); say "foo".Stringy # should this loop indefinitely?
m: class SQLString { }; my $stringy = Str.^find_method("Stringy"); my $handler = $stringy.wrap(method () { SQLString.new(:str(callsame)) }); say "foo".Stringy # should this loop indefinitely? 12:59
camelia rakudo-moar c3c86d: OUTPUT«(timeout)»
timotimo Stringy has always had a tendency to cause endless loops
lizmat perhaps we need to look at what you're trying to achieve ? 13:00
dogbert17 nothing :) took it from a MoarVM bug report, it used to crash before jnthn made a callsame fix 13:01
if the snippet is supposed to loop endlessly I think the original bug can be cinsidered fixed 13:02
*considered
it's from github.com/MoarVM/MoarVM/issues/412 13:05
lizmat www.bleepingcomputer.com/news/secu...pril-2017/ # does this affect our use of the JVM backend ? 13:18
timotimo i'm not sure we sign our .jar files at all? 13:22
lizmat has no idea 13:24
timotimo but i expect it's a trivial matter to use a different hash there. like, changing a commandline flag or something 13:25
Geth kudo/nom: d71bf1e90c | (Elizabeth Mattijsen)++ | src/core/Supply.pm
Implement Supply.skip()
 13:49
moritz stackoverflow.com/questions/4179160...-different that looks like a bug to me 14:42
timotimo moritz: i think you'd have to use "role Either is Alpha is Digit" 14:43
moritz timotimo: why? then you lose all the advantages that role semantics bring 14:45
timotimo i just think you have to use derivation rather than composition?
but ... *shrug*
not a thing i've been doing so far
brokenchicken m: say (Empty xx 4)[0] 18:16
camelia rakudo-moar d71bf1: OUTPUT«()␤»
brokenchicken m: dd (Empty xx 4)[0]
camelia rakudo-moar d71bf1: OUTPUT«Empty␤»
brokenchicken m: dd (Slip.new() xx 4)[0]
camelia rakudo-moar d71bf1: OUTPUT«Nil␤»
brokenchicken m: dd (Slip.new(42, 44) xx 4)[0] 18:17
camelia rakudo-moar d71bf1: OUTPUT«Int <element> = 42␤»
brokenchicken Seems Empty is being mishandled here?
:S make t/spec/S16-io/newline.t fails but prove -e './perl6' -vlr t/spec/S16-io/newline.t succeeds 18:26
and t/spec/S10-packages/precompilation.t consistently fails
mm.. and same thing with t/spec/integration/advent2011-day03.t make fails but prove succeeds :S
brokenchicken git resets and tries another run 18:27
ah, k. Prolly something with precomp. My VM froze mid one spectest run and I just ran the second one. I now rebuilt and all seems peachy 18:36
MasterDuke i just cleaned out the .precomp directory in my rakudo/t/spec/packages earlier this morning, it was >400Mb, with files/directories >6m old 18:38
brokenchicken yeah, once I ran `rm -fr rakudo` and it ran for several minutes... 18:45
Geth kudo/nom: 4e49ec135d | (Zoffix Znet)++ | src/core/List.pm
Fix infix:<xx>(..., *) for empty Slips

The current code makes the assumption that when a Slip is encountered, it is non-empty, and it saves it in an Array and shifts from that array, to obtain next value. This causes a Failure when the Slip is empty and there is nothing to shift.
Fix by checking whether we can shift and returning IterationEnd if can't.
Fixes RT#130619: rt.perl.org/Ticket/Display.html?id=130619		 18:50
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130619
ast: 75e15bbb2a | (Zoffix Znet)++ | S03-operators/repeat.t
Test empty slip with xx * works

RT#130619: rt.perl.org/Ticket/Display.html?id=130619 Rakudo fix: github.com/rakudo/rakudo/commit/4e49ec135d		 18:53
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130619
brokenchicken m: dd (Empty.new xx 4)[0] 18:58
camelia rakudo-moar d71bf1: OUTPUT«Nil␤»
brokenchicken m: dd (Empty xx 4)[0]
camelia rakudo-moar d71bf1: OUTPUT«Empty␤»
brokenchicken (one's thunked the other isn't)
prolly makes sense... 18:59
Geth kudo/nom: 15753fd3c3 | (Elizabeth Mattijsen)++ | src/core/Supply.pm
Fix scoping issue with Supply.skip, spotted by jnthn++
 19:11
ast: 943b3ad515 | (Elizabeth Mattijsen)++ | S32-list/skip.t
Add tests for Any.skip
 19:34
kudo/nom: cbb7130038 | (Elizabeth Mattijsen)++ | t/spectest.data
Make sure we run Any.skip tests
 19:37
ast: 24af69c6ca | (Elizabeth Mattijsen)++ | S17-supply/skip.t
Add basic Supply.skip tests
 19:41
kudo/nom: 8a448d4c63 | (Elizabeth Mattijsen)++ | t/spectest.data
Make sure we run Supply.skip tests
 19:42
kudo/nom: e0201f15df | (Elizabeth Mattijsen)++ | src/core/Any-iterable-methods.pm
Introducing Any.batch(n)

Since Supply has a .batch, I figured we should probably also have an Any.batch, just to keep the balance :-)		 19:51
ast: ca42a37505 | (Elizabeth Mattijsen)++ | S32-list/batch.t
Add some basic Any.batch tests
 20:07
kudo/nom: ce76f4c5a0 | (Elizabeth Mattijsen)++ | t/spectest.data
Make sure we run the Any.batch tests
MasterDuke is anyone else getting `Argument "3.36_01" isn't numeric in numeric gt (>) at t/harness5 line 118.` when running `make m-spectest`? 20:49
This is perl 5, version 24, subversion 1 (v5.24.1) built for x86_64-linux-thread-multi 20:50
moritz rebuilds rakudo to reproduce 20:52
MasterDuke i don't think it's a rakudo thing 20:55
pretty sure perl5 was updated on my machine a day or two ago
Geth kudo: MasterDuke17++ created pull request #1003:
Speedup postincrement and postdecrement by ~5%		 20:56
kudo/nom: fd74c91791 | (Daniel Green)++ | src/core/Int.pm
Speedup postincrement and postdecrement by ~5%

Change an assign into a bind.		 20:57
kudo/nom: 8729048c3f | lizmat++ | src/core/Int.pm
Merge pull request #1003 from MasterDuke17/speedup_postinc_and_postdec

Speedup postincrement and postdecrement by ~5%
MasterDuke lizmat++ that was fast 20:58
lizmat I actually had been looking at that one just minutes before your PR :-)
MasterDuke heh. i also tried wrapping the lines in a nqp::stmts(), but i measured no difference 20:59
lizmat did you try doing the binding of b inside the nqp::add_i ? 21:00
MasterDuke nope 21:01
lizmat that *may* help as well :-)
it's something I would try :-) 21:02
MasterDuke i'll run some more tests
lizmat MasterDuke++
Geth kudo/nom: 3a4b26283b | (Zoffix Znet)++ | t/harness5
Avoid warning when TAP::Harness's version ain't numeric
 21:09
MasterDuke lizmat: no noticeable difference 21:26
lizmat ok, I guess the optimizer / spesh / JIT are doing their work then :-)
MasterDuke in rakudo/src there are 158 `{*}` and 337 `{ * }`. anybody have strong opinions about whether we should be consistent? and if so, which way? 21:36
brokenchicken has no opinion 21:37
timotimo fewer spaces to make parsing faster!!!!!!k ;)
brokenchicken :)
Geth kudo/nom: 0fdb9e4e40 | (Elizabeth Mattijsen)++ | 2 files
nativeshaped.pairs returns Pair.value that is a LHS

Before this, (my int @a[2;2]).pairs[0].value++ would fail with:
   Cannot resolve caller postfix:<++>(Int)		 21:38
MasterDuke not entirely a bad idea: gist.github.com/MasterDuke17/77230...lts-csv-L8
timotimo We can make this file beautiful and searchable if this error is corrected: It looks like row 97 should actually have 1 column, instead of 2. 21:39
oh?
MasterDuke _ws and ws are #8 and #18 by exclusive time for profile-compile of a rakudo build
github thinks it's a csv, so the infix<,> messes it up 21:40
timotimo ah, hehehe
pretty good
MasterDuke is partially to blame for naming the file .csv
timotimo tsv would be better, because i'm pretty sure we don't have an operator \t so far :D 21:41
moritz not YET :-)
timotimo is probably possible to put in already 21:42
RabidGravy I keep meaning to post this with regard to the performance thing as it seemed pertinent jamesclear.com/marginal-gains 21:47
yoleaux2 21 Jan 2017 21:35Z <MasterDuke> RabidGravy: yep, github.com/MoarVM/MoarVM/pull/501
RabidGravy MasterDuke++ # nice one 21:48
Geth kudo/nom: 1181afec40 | (Elizabeth Mattijsen)++ | 2 files
Make sure my int @a[2].kv gives LHS valueo

Before this, (my int @a[2]).kv[1]++ would fail with:
   Cannot resolve caller postfix:<++>(Int)		 21:49
kudo/nom: e195e5f990 | (Elizabeth Mattijsen)++ | 2 files
Make sure my int @a[2].values gives LHS values

Before this, (my int @a[2]).values[0]++ would fail with:
   Cannot resolve caller postfix:<++>(Int)		 22:04
lizmat m: my int @a[3] = (0,1,2); @a = (); dd @a # perhaps unexpected behaviour of 1dim shaped arrays ? 22:31
camelia rakudo-moar e195e5: OUTPUT«array[int].new(:shape(3,), [0, 1, 2])␤»
lizmat m: my int @a = (0,1,2); @a = (); dd @a # perhaps unexpected behaviour of 1dim shaped arrays ? 22:32
camelia rakudo-moar e195e5: OUTPUT«array[int].new()␤»
timotimo m: my int @a[3] = 0, 1, 2; @a = (1,); dd @a
camelia rakudo-moar e195e5: OUTPUT«array[int].new(:shape(3,), [1, 1, 2])␤»
timotimo at least it's consistent with that?
lizmat yes, it is 22:33
but I'm thinking, suppose we pass an array to a module somewhere, and that module does = () thinking it initializes it
m: sub a(@a) { @a = () }; a(my @a[3] = (42,66,666)); dd @a 22:34
camelia rakudo-moar e195e5: OUTPUT«Array.new(:shape(3,), [42, 66, 666])␤»
lizmat m: sub a(@a) { @a = () }; a(my @a = (42,66,666)); dd @a 22:35
camelia rakudo-moar e195e5: OUTPUT«Array @a = []␤»
lizmat feels a bit like a footgun to me
unless we can actually have something in a signature to match shaped arrays 22:36
I came across this while investigating the native shaped array coverage, which is minimal atm 22:37
timotimo m: my int @a[3] = 9, 9, 9; @a = Empty; dd @a
camelia rakudo-moar e195e5: OUTPUT«array[int].new(:shape(3,), [9, 9, 9])␤»
timotimo to me, that's more footgunny, as it explicitly says "empty"
lizmat yup
timotimo but i really need sleep now
lizmat timotimo: good night, sleep tight!
timotimo i'll do my best! :)
jnthn At the same time, `@shaped-array = @replace-first-elems-with-this` working fine then exploding when the second array contains zero elems would be a nasty discontinuity. 23:13
Not sure we should break a useful thing's boundary case for the sake of a footgun. 23:14
lizmat well, I'm working on adding a warning, would that be ok ?
jnthn Not really :( 23:15
lizmat also for the () case ?
jnthn I mean, I guess you could put quietly but it's still a discontinuity
Can we distinguish Empty?
I can go with warning on Empty
m: say |()
camelia rakudo-moar e195e5: OUTPUT«␤»
jnthn m: say (|()).perl
camelia rakudo-moar e195e5: OUTPUT«slip()␤»
jnthn OK, good, we can distinguish that from Empty :)
I'm fine with the warning on Empty 23:16
lizmat ok
but then I will need to rethink the code and warning
and that will be tomorrow :-)
jnthn Yeah, I'm gonna turn in too :) 23:17
busy day for me tomorrow; likely won't be around all that much.
lizmat ok, sleep well!
jnthn Thanks, you also :) 23:18
'night, #perl6-dev
MasterDuke m: say sprintf '%0۳d', -3.1415 23:48
camelia rakudo-moar e195e5: OUTPUT«Directive d not applicable for type Rat␤␤»
MasterDuke m: say sprintf '%0۳d', -31415
camelia rakudo-moar e195e5: OUTPUT«Directive d not applicable for type Int␤␤»
MasterDuke m: say sprintf '%03d', -31415
camelia rakudo-moar e195e5: OUTPUT«-31415␤»
MasterDuke m: say sprintf '%۳.3Gd', -3.1415 23:49
camelia rakudo-moar e195e5: OUTPUT«This representation (Null) does not support elems (for type VMNull)␤␤»
MasterDuke m: say sprintf '%3.3Gd', -3.1415
camelia rakudo-moar e195e5: OUTPUT«-3.14d␤»
MasterDuke m: say sprintf '%3.۳G', -3.1415
camelia rakudo-moar e195e5: OUTPUT«Directive g not applicable for type Rat␤␤»
MasterDuke m: say sprintf '%3.3G', -3.1415
camelia rakudo-moar e195e5: OUTPUT«-3.14␤»
MasterDuke m: say sprintf '%۳.3G', -3.1415 23:50
camelia rakudo-moar e195e5: OUTPUT«This representation (Null) does not support elems (for type VMNull)␤␤»
MasterDuke .ask TimToady do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work? 23:52
yoleaux2 MasterDuke: I'll pass your message to TimToady.
MasterDuke .ask jnthn do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work?
yoleaux2 MasterDuke: I'll pass your message to jnthn.
MasterDuke .ask lizmat do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work?
yoleaux2 MasterDuke: I'll pass your message to lizmat.
MasterDuke .ask samcv do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work?
yoleaux2 MasterDuke: I'll pass your message to samcv.
MasterDuke .ask bartolin do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work? 23:53
yoleaux2 MasterDuke: I'll pass your message to bartolin.
samcv MasterDuke, that would be cool 23:54
yoleaux2 23:52Z <MasterDuke> samcv: do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work?
samcv let me poke a few things and see if I have any other comments 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!