»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
00:00 autarch1 left
AlexDaniel Juerd: ok I think that there's an easier way to confirm your solution… 00:00
I'll just slam random input into your solution and into mine, let's see if it finds any bugs… 00:01
00:11 musiKk left
AlexDaniel m: say ‘abc’ eq ‘abc’ eq ‘abc’ eq ‘abc’ 00:11
camelia rakudo-moar 8fdaad: OUTPUT«True␤»
AlexDaniel m: my @a = ‘abc’, ‘abc’, ‘abc’, ‘abc’; say [eq] @a 00:12
camelia rakudo-moar 8fdaad: OUTPUT«True␤»
00:13 autarch joined 00:16 poele joined
AlexDaniel Juerd: yeah! 00:23
Juerd: gist.github.com/AlexDaniel/791f4994d35e2d8a63b1 00:24
Juerd AlexDaniel: What the... 00:27
AlexDaniel ya :)
Juerd AlexDaniel: That's not even valid input, though :) 00:28
That can never happen in a game
AlexDaniel errr, are you sure?
I mean *exactly* this input is not going to happen
Juerd Yes, games always grow from the center.
AlexDaniel but a similar situation probably can
Juerd Show me :) 00:29
(Regardless, I have no idea what's happening here!)
Oh, 8 requires special treatment :( 00:30
00:31 poele left
Juerd That is, for the edge 00:31
I think I'll leave it at the current state
Feel free to debug if you do like to
00:31 damnlie left
Hotkeys what's the best way to find the next existing number in an array given a number that doesn't exist 00:32
00:32 damnlie joined
Hotkeys eg 00:33
given @a = [1, 7, 9, 10] and $n = 2, how would I get 7?
oh wait
m: @a.first(* 00:34
camelia rakudo-moar 8fdaad: OUTPUT«5===SORRY!5=== Error while compiling /tmp/7Gn1ss7sg7␤Unable to parse expression in argument list; couldn't find final ')' ␤at /tmp/7Gn1ss7sg7:1␤------> 3@a.first(*7⏏5<EOL>␤»
Hotkeys er
m: my @a = [1, 7, 9, 10]; my $n = 2; @a.first(* > $n).say
camelia rakudo-moar 8fdaad: OUTPUT«7␤»
AlexDaniel Juerd: this position is just fine
Hotkeys okay got it
AlexDaniel Juerd: not sure if I did everything right but something like this: gist.github.com/AlexDaniel/65623bce3cd1f360eed0 00:35
Juerd AlexDaniel: I lost interest though 00:36
AlexDaniel XD
Juerd AlexDaniel: I realised that I'm spending way too much time on this.
00:36 addison joined
AlexDaniel true-true 00:36
Juerd: I think that this is the last bug though 00:37
timotimo sdh: sorry, i haven't caught up with recent JVM work yet
synopsebot6 seems happier now that it has no limit on number of processes it may spaws 00:39
spawn
00:40 ely-se left
AlexDaniel Juerd: actually, yes, it is the last bug. I'm currently running both solutions and comparing the output. After like 2 minutes it found nothing 00:40
Juerd: that previous bug was actually found by this script… but I think that I got lucky
Juerd Just find it then :P
timotimo grondilu: with recent changes taken into account, my code for white noise is now at something like 20 FPS, how does that sound to you? :)
00:42 cfedde left 00:45 gfldex left 00:47 gfldex joined
Juerd AlexDaniel: Out of the 6 that my solution found, I don't know which one is wrong or missing. 00:48
mspo i.imgur.com/mRNOELn.png
AlexDaniel mspo: awesome
mspo AlexDaniel: I've never seen anything so relevant to this channel ;)
AlexDaniel I've never seen these quotes in Estonia though… 00:49
Juerd: hmm did you just fix something? 00:50
00:51 cfedde joined
AlexDaniel because it looks like it went away 00:51
Juerd: gist.github.com/AlexDaniel/2b23299c8b68fa4a7426 00:54
Juerd: this one is slightly different 00:55
Juerd: seems like it cannot find any solution from the edge: gist.github.com/AlexDaniel/2b23299c8b68fa4a7426 00:57
Juerd: “Out of the 6 that my solution found” – this was not a test case, it was just a proof that this state is valid 00:58
“proof” because I played it by hand
mspo: any stats about which one is the most common? 01:00
mspo AlexDaniel: just a link I saw on reddit, not much background info 01:02
01:03 cfedde left 01:04 cfedde joined 01:08 cfedde left 01:10 cfedde joined 01:26 addison left 01:27 kid51 joined 01:28 pierrot left 01:34 pierrot joined 01:37 pierrot left 01:42 Ben_Goldberg joined 01:44 BenGoldberg_ left 02:00 PerlJam left 02:01 PerlJam joined 02:02 sortiz joined
sortiz \o #perl6 02:02
02:03 lichtkind left
skids \o 02:05
masak o/ 02:06
I'm toying around with LF and CRLF. I observe the following: if I read a file with CRLF line endings with Rakudo, and just print it back out with -n or -p, it comes out with LF line endings. (I'm on Linux.) 02:09
this, I think is a good default.
I haven't yet been able to override that default. I don't think it's possible with today's Rakudo. 02:10
Hotkeys is there an option for as-is yet
masak S28 says there's supposed to be a $*OUT.output-record-separator, but there isn't in Rakudo right now 02:11
Hotkeys so you don't convert any of the line endings
and just leave them as is
skids .nl-out?
masak skids: see above; S28 says it's called $*OUT.output-record-separator
I'm a little confused as to why the naming between $*IN.input-line-separator() and $*OUT.output-record-separator() is inconsistent 02:12
skids Yes but S28 is probably older than the whole CRLF thing right before 6.c
masak skids: yes; that's part of why I'm asking
m: say $*OUT.nl-out
camelia rakudo-moar 8fdaad: OUTPUT«␤␤»
masak oh, so that one does exist.
masak tries
it's confusing when S28 says one thing and Rakudo implements another ;) 02:13
skids Yeah we need a re-whirlpool-the-design-docs initiative 02:14
masak ok, .nl-out works 02:15
I don't know what "re-whirlpool" means, but it seems to me it'd be worth it to take S28 as a starting point when introducing new replacements for old Perl 5 special variables :) 02:17
02:22 addison joined
sortiz For "re-whirlpool" I understand that, now that roasts tests are somewhat normative, when there are inconsistencies between them and S*, the tested form should be taken to S*. 02:23
masak t/spec/S32-num/power.rakudo.moar
TODO passed: 13-15, 68-70
02:24 Ben_Goldberg is now known as BenGoldberg
masak sortiz: the tests have been normative in that sense for years, IMO. 02:24
maybe even since they were created.
timotimo fair enough, but development reality made the specs get outdated at a much faster pace than they used to
masak that doesn't seem to be the problem here, though 02:25
the problem is that S28 was never even consulted
sortiz That is what I know, yes.
masak I'm just curious why that is
timotimo i think the names used to be implemented that way in rakudo, but were later changed again? 02:26
hm, no, that's not it
02:27 addison left 02:28 BenGoldberg left 02:29 BenGoldberg joined, BenGoldberg left 02:30 BenGoldberg joined
masak no, there used to be a single .nl, and then they were split up as .nl-in and .nl-out 02:32
sortiz In that sense what was missing at the time was to update the specs. 02:37
masak or to read them 02:38
geekosaur that was near the start of the final rush toward 6.c, no? 02:39
think people were in too much of a hurry
skids IIRC that change was done at a pretty frantic time. In addition to that, lizmat++ spent a lot of time reworking IO design docs, kept soliciting feedback or for some sort of approval for a merge, but was feeling a bit warnocked on the matter. Not that nl-out was part of that. 02:40
masak yes, yes 02:41
sortiz Btw, in roast I only found two mentions to 'nl-out' and only in the context of exceptions. 02:42
skids But it was reasonable with the general sense that IO needed a rework to treat that part of the design as a bit more fungeable than the rest.
masak it's far too easy to come in after a change and expect a full accounding of events and why things were done the way they were
I'm not trying to do that
I'm just interested in some sort of consistency going forward
timotimo is likely to go forward to bed in the very near future
masak S28 is among the less stable/reliable documents, I'm aware of that
but it still has value IMO
02:45 ilbot3 left, Herby_ joined
skids S28 is not where I would have thought to check for that :-). Anyway So far at least we have some rules for what can change in roast 6.c-errata. I don't know if that's the branch that should be used to update design docs but it should be relatively stable. 02:45
Herby_ Evening, everyone!
yoleaux 20 Feb 2016 16:46Z <timotimo> Herby_: seems like nobody pointed this out to you, but there's already an irc bot framework in the ecosystem :) it's what synopsebot is written with
masak we seem to have spectests for .nl-in but not for (the behavior of) .nl-out 02:46
02:47 ilbot3 joined
skids Yeah unfortunately (for this purpose) the rules for 6.c-errata are "no new tests, just unfudges" and unless I missed it there's no level between master and that yet. 02:48
masak e20a16b2bea9eebaab2a05496e8ff134a4cde9f2 from roast is interesting. 02:49
there used to be a test with `$RT123888.input-line-separator`, but it got changed to `$RT123888.nl`
the commit comment seems unrelated to that change. 02:50
I can see how no-one thought to update S28 if that was the only test, and if the change was made in passing without acknowledging it in the commit comment ;)
02:54 zpmorgan left
timotimo good night #perl6 02:55
skids o/
Herby_ \o
skids
.oO(Herby_ always waves with the opposite hand from me. Contrarian. :-)
 02:56
Herby_ it is my nature :)
02:57 kid51 left
AlexDaniel what is FSA? 03:00
masak AlexDaniel: "finite state automaton"?
ugexe fixed size allocator 03:01
masak Federal Student Aid 03:02
Full Speed Ahead!
sortiz AlexDaniel, In MoarVM The fixed size allocator provides a thread-safe mechanism for getting and releasing fixed-size chunks of memory." 03:09
AlexDaniel thanks 03:13
03:18 sdh left
dalek ast: 04784c1 | skids++ | S01-perl-5-integration/context.t:
Fix some spelling
 03:50
ast: 2f4cda8 | skids++ | S01-perl-5-integration/context.t:
Missed some spelling to fix
ast: e57bfe9 | skids++ | S01-perl-5-integration/context.t:
Merge pull request #105 from skids/master

Did this as a PR to prove to myself it does not merge automatically
skids Good. I must have been confused the first time I tried that. 03:51
03:53 cpage_ left 03:55 noganex_ joined
dalek kudo/nom: 5dcf96f | (Carl Masak)++ | src/core/IO.pm:
remove predeclaration of Instant in core/IO.pm

This one hasn't been needed since 509545a46136390feef5f3c0f56943ccf6a1881a.		 03:58
03:59 noganex left 04:04 sunnavy left 04:07 sunnavy joined, yqt left
MadcapJake . 04:11
yoleaux 20 Feb 2016 21:45Z <pmurias> MadcapJake: lab name => "common case", try => {...}; lab name => "complex case", try => {variant1 => {...}, variant2 => {...}}
04:15 Ben_Goldberg joined 04:16 BenGoldberg left 04:19 Actualeyes joined 04:27 mattp_ joined 04:34 cognominal left, cognominal joined
Ben_Goldberg m: use POSIX; 04:42
camelia rakudo-moar 5dcf96: OUTPUT«===SORRY!===␤Could not find POSIX in:␤ /home/camelia/.perl6/2015.12-403-g5dcf96f␤ /home/camelia/rakudo-m-inst-1/share/perl6/site␤ /home/camelia/rakudo-m-inst-1/share/perl6/vendor␤ /home/camelia/rakudo-m-inst-1/share/perl6␤ CompUn…»
04:43 Sgeo__ joined 04:44 kaare_ joined 04:45 Sgeo_ left 04:47 grassass joined 04:49 edenc left 04:51 kaare_ left, edenc joined, kaare_ joined 04:52 AndyDee joined, adrusi left 04:54 kst` joined 04:55 kst left 04:56 adrusi joined, Ben_Goldberg left 04:58 vendethiel joined 05:21 vendethiel left 05:34 Cabanossi left 05:37 Cabanossi joined 05:43 rburkholder left 05:48 adu joined 06:04 khw left 06:13 davido_ joined
jdv79 is panda bootstrap broken? 06:21
06:26 Herby_ left 06:27 autarch left 06:44 cpage_ joined 06:46 CIAvash joined
sortiz jdv79, I didn't see any problem in my last 'rakudobrew build' 06:50
06:55 alpha123 left, AndyDee left
jdv79 yeah its fine 06:57
having a env issue which is causing the wrong perl6 to be run
masak .oO( env: the silent killer ) 07:04
jdv79 anyone know why moving ~/.rakudobrew would stop its PATH entries from magically appearing
NO. there's .profile and i was looking for .bash* files. :( 07:06
07:08 aindilis joined 07:57 adu left 08:09 skids left 08:10 davido_ left 08:11 pmurias joined 08:12 rindolf joined 08:13 darutoko joined 08:16 pmurias left 08:30 sortiz left 08:41 DoverMo joined
DoverMo my nib broke 08:41
Hotkeys unbreak it 08:45
DoverMo: what happened?
AlexDaniel break it again, try to figure out why exactly it breaks. Then submit a bug report. 08:47
08:50 DoverMoo joined, DoverMoo left 08:53 DoverMo left 08:59 chienjo joined 09:00 chienjo left 09:02 _mg_ joined 09:12 psch joined
psch hi #perl6 o/ 09:14
.seen sdh
yoleaux I saw sdh 20 Feb 2016 23:45Z in #perl6: <sdh> Ah, thanks. Is it in the works? I was digging around the runtimes to see what I could hack together, but didn't have any luck.
psch ah, doesn't tell when someone left...
.tell sdh github.com/perl6/nqp/tree/standalo.../tools/jvm has some probing from a while ago wrt deploying jars. i'm unsure in how far it still works nowadays, though. 09:15
yoleaux psch: I'll pass your message to sdh.
09:16 wamba joined 09:20 molaf joined 09:23 ely-se joined, grassass left
dalek kudo/nom: dae9f6a | okaoka++ | LICENSE:
Fix a license duration 2000-2015 -> 2000-2016
 09:25
kudo/nom: 7010f32 | niner++ | LICENSE:
Merge pull request #717 from okaoka/fix-license-duration

Fix a license duration 2000-2015 -> 2000-2016
09:25 iH2O joined 09:29 agentzh left 09:31 petercommand left 09:32 petercommand joined 09:35 TEttinger left 09:37 RabidGravy joined 09:38 jeek left 09:42 agentzh joined, agentzh left, agentzh joined 09:50 iH2O left 09:51 vendethiel joined
RabidGravy HARR! 09:54
10:08 _mg_ left
masak H∀RR! 10:09
RabidGravy W€€! 10:11
[Tux] test 22.532
test-t 11.956
csv-parser 49.302
RabidGravy Oöøh! nice 10:12
started going down again
10:13 ely-se left
RabidGravy if I have a parametric role, and I compose that into another role is there a way of applying the parameter when I apply the resulting role to an actual class 10:15
a bit like
m: role Foo[$bar?] { method bar() { say $bar }}; role Bar does Foo {}; my $a = "yu"; $a does Bar["yuck"]; $a.bar
camelia rakudo-moar 7010f3: OUTPUT«No appropriate parametric role variant available for 'Bar'␤ in any specialize at gen/moar/m-Metamodel.nqp line 2600␤ in any specialize at gen/moar/m-Metamodel.nqp line 2194␤ in any compose at gen/moar/m-Metamodel.nqp line 2979␤ in any generate_…»
RabidGravy but working obviously
10:15 vendethiel left
masak I fail to see why that shouldn't work 10:16
RabidGravy well I guess because Bar itself doesn't have parameters, 10:17
10:19 TEttinger joined
arnsholt role Bar[$b] does Foo[$a] { ... } perhaps? 10:23
s/$a/$b/
Obviously
10:28 CIAvash left 10:29 gregf left
RabidGravy Nope, I'll have to stick with the yucky I have at the moment 10:32
10:36 AlexDaniel left 10:37 spider-mario joined 10:40 renormalist joined 10:41 _dolmen_ joined 10:44 CIAvash joined
RabidGravy also need more coffee 11:00
nine coffee++ 11:02
11:10 hankache joined
hankache hola #perl6 11:10
11:14 _dolmen_ left 11:25 apiw joined 11:32 domidumont joined 11:39 vendethiel joined 11:45 musiKk joined 11:49 apiw left 11:50 CIAvash left 11:56 apiw joined 11:58 hankache left 12:01 vendethiel left 12:15 vendethiel joined 12:19 FROGGS joined 12:34 apiw left 12:35 rburkholder joined 12:36 vendethiel left 12:40 kid51 joined 12:48 _mg_ joined 12:51 musiKk left 12:52 FROGGS left 12:53 FROGGS joined 12:54 Skarsnik joined 12:55 pmurias joined 12:58 zpmorgan joined
pmurias .tell do we need %Test::Lab::context for anything? in ruby the defaults are set per class, having that global doesn't seem to give us much 12:58
yoleaux pmurias: I'll pass your message to do.
13:01 BenGoldberg joined 13:09 aigan_ left
Skarsnik Hello 13:16
pmurias hi
RabidGravy erp 13:18
13:19 Relsak joined
FROGGS pmurias: was it intentional that you .told "do" something? 13:20
13:27 apiw joined
pmurias .tell MadcapJake do we need %Test::Lab::context for anything? in ruby the defaults are set per class, having that global doesn't seem to give us much 13:30
yoleaux pmurias: I'll pass your message to MadcapJake.
pmurias FROGGS: no, thanks
any ideas when the rakudo-js grant vote will take place? 13:32
nine Does rakudo always use the Backtrace class when printing backtraces? 13:35
13:36 risou is now known as risou_awy, risou_awy is now known as risou
jnthn nine: afaik, yes, except when you pass --ll-exception, then we just let the VM spew out whatever it will 13:36
pmurias do we want to add column info to the Backtrace class? 13:37
nine jnthn: ah, that could explain it, thanks!
I'm working on getting rid of absolute paths in precomp files. I already turned them into paths relative to the repository that contains the files and now "just" need to assemble it back together on output
jnthn nine: Will this rid us of the sha-1s in backtraces too? 13:38
nine And out of lazyness I tested with NativeCall and --ll-exception :)
jnthn: yes, I already add the requested module's name in () to the displayed source path
at perl#sources/24DD121B5B4774C04A7084827BFAD92199756E03 (NativeCall):261 13:39
jnthn nine: Is there a way we can show the path the source file had inside its distribution?
Instead of the SHA-1
13:40 vendethiel joined
Skarsnik hm, for rt.perl.org/Ticket/Display.html?id=127345 it's in the Backtrace class? 13:41
nine I'd rather like to keep the SHA-1 in the output as you can copy&paste this path to look at the source. Adding the module name to the output should already make the backtrace much more readable
13:41 Begi joined 13:51 Psyche^ joined
Begi is there a method to keep only letters in a string ? 13:53
without spaces and ponctuation
jnthn m: say "oh! I've been stripped!".subst(/\W+/, '', :g) 13:54
camelia rakudo-moar 7010f3: OUTPUT«ohIvebeenstripped␤»
jnthn Though that'll keep _ too
13:55 Psyche^_ left
jnthn m: say "o__h! I've been stripped!".subst(/<-:Letter>+/, '', :g) 13:55
camelia rakudo-moar 7010f3: OUTPUT«ohIvebeenstripped␤»
jnthn That used the Unicode Letter property
m: say "o__h! I've been stripped!".subst(/<-:L>+/, '', :g)
camelia rakudo-moar 7010f3: OUTPUT«ohIvebeenstripped␤»
jnthn Can shorten it to that if golfing :) 13:56
Ulti www.lamdu.org/ this feels like something you could maybe do with Rakudo 13:59
14:00 musiKk joined
Begi Thanks jnthn, it works ! I'm reading the doc about this method 14:01
14:01 apiw left
moritz m: say "o__h! I've been stripped!".comb(/<:L>+/).join 14:03
camelia rakudo-moar 7010f3: OUTPUT«ohIvebeenstripped␤»
lizmat sometimes I think .comb should also be able to take a Callable, like .grep 14:08
BenGoldberg m: say [ "o__h! I've been stripped!".comb(/<:L>+/) ] 14:10
camelia rakudo-moar 7010f3: OUTPUT«[o h I ve been stripped]␤»
14:10 musiKk left 14:13 Amendil joined
BenGoldberg What would be the semantics, if .comb *did* take a Callable? 14:14
moritz lizmat: you can jsut write .comb.grep
lizmat yeah, but I was thinking performance 14:15
try this for difference: .comb(/../) and .comb(2) :-)
14:16 apiw joined
jnthn I think we'd get more mileage out of making simple regexes like that a bunch faster. :) 14:17
14:19 autarch joined
lizmat jnthn: true, but Callable has it's own rewards in that respect... and we don't really have a .grep on characters 14:19
jnthn lizmat: I don't know it's worth a lot over .comb.grep tbh 14:20
nine I don't get it. Why do I get "Could not find symbol '&RepositoryRegistry'" for a $file = CompUnit::RepositoryRegistry.repository-for-name(... in Backtrace's AT-POS at runtime?
moritz lizmat: the approach of adding ever more specialized built-ins for performance doesn't scale
lizmat moritz: I said "sometimes I think" :-) 14:23
14:23 vendethiel left
moritz lizmat: noted :-) 14:23
jnthn The trade-off of note is that every one we add increases base memory 14:24
nine Ah, predeclaring CompUnit::RepositoryRegistry fixes it
moritz nine: bootstrapping is fun :-)
14:25 beatdown joined 14:27 _mg_ left
dalek c: dafabe7 | (Simon Ruderich)++ | doc/Type/Hash.pod:
Hash: fix display of code listing

The tab caused the listing to be displayed as multiple listing separated by an empty line.		 14:29
14:34 dududuedu joined 14:38 stux|RC-only left 14:40 vendethiel joined 14:41 stux|RC-only joined
rudi_s I'm looking at Hash documentation and I'm a little confused about the last example in "Loop over hash keys and values": for %vowels.kv { .say } with "This would print the list one line at a time". <- This is not true, it prints all keys and values. Any idea how the code is supposed to look? 14:43
And in the example before that, should we mention .flat as an alternative way to get the usual -> $k, $v { .. } way? 14:44
14:45 Zero_Dogg left
moritz +1 to the latter 14:46
FROGGS rudi_s: for the first snipped, it should probably say "This would print one Pair at a time."
psch m: my %h = a => 0, b => 1, c => 3; for %h.kv { .say }
camelia rakudo-moar 7010f3: OUTPUT«a␤0␤c␤3␤b␤1␤»
FROGGS ohh
of course :o) 14:47
the sentence is still weird though
psch yeah, "the list" seems insufficiently explained
FROGGS yeah
moritz feel free to replace it or toss it
14:47 Ben_Goldberg joined 14:48 apiw left
lizmat m: sub a(*@a) { state $b = -> { say @a.shift }; $b() }; a(1,2,3); a(4,5,6) # jnthn moritz: is this a scoping issue or not? 14:48
camelia rakudo-moar 7010f3: OUTPUT«1␤2␤»
lizmat I would expected 1,4 there
moritz lizmat: looks fine to me
lizmat: the state initializer is only run the first time, and the pointy block is a closure 14:49
14:50 BenGoldberg left
rudi_s I'd drop the code example including the last sentence of the paragraph as I've no idea what it's supposed to say, any objections? 14:50
lizmat hmmm...
moritz rudi_s: it's supposed to give you an idea how iterating a hash works
psch rudi_s: fwiw, explaining what actually happens is probably sensible, i'm at a lost how to phrase it though
rudi_s Should I change the previous example to ust .flat per default and mention that without .flat the pointy block needs to split the argument. 14:51
moritz what happens when you iterate over %h, %h.kv, %h.pairs, %h.keys etc?
rudi_s: nah, just offer both; There is more than one way to do it!
rudi_s moritz: But that's the first example in the section, I'm talking about the last one.
psch "This would print the list consisting of unordered and flattened Pairs one element at a time." seems slightly wrong...
rudi_s moritz: Ok.
moritz ah well, if it's redundant and unclear, by all means, drop it 14:52
rudi_s "The output of this operation is a L<Seq>," <- this seems to be wrong too, the result is a List, not a Seq.
psch m: my %h = a => 0, b => 1, c => 3; say %h.kv.WHAT
camelia rakudo-moar 7010f3: OUTPUT«(Seq)␤»
rudi_s m: my %h = a => 0, b => 1, c => 3; say %h.sort(*.key)>>.kv.WHAT 14:53
camelia rakudo-moar 7010f3: OUTPUT«(List)␤»
rudi_s m: my %h = a => 0, b => 1, c => 3; say %h.sort(*.key)>>.kv[0].WHAT 14:54
camelia rakudo-moar 7010f3: OUTPUT«(List)␤»
14:54 BenGoldberg_ joined
moritz .kv gives you Seq of List objects, no? 14:54
psch huh
isn't that calling .kv on Pairs?
m: my %h = a => 0, b => 1, c => 3; say %h.sort(*.key)>>.WHAT 14:55
camelia rakudo-moar 7010f3: OUTPUT«(List)␤»
psch err, WHAT *shakes fist*
moritz m: say %*ENV.sort[0].^name
camelia rakudo-moar 7010f3: OUTPUT«Pair␤»
psch m: my %h = a => 0, b => 1, c => 3; say %h.sort(*.key)>>[0].WHAT
camelia rakudo-moar 7010f3: OUTPUT«(List)␤»
14:55 BenGoldberg_ left
psch right, so the sorted Pairs each get a .kv, which is why the order remains 14:55
rudi_s moritz: But not with >>.kv, not ide awhy.
14:56 Ben_Goldberg left
moritz because >> is magical about the way how far it descends into a nested data structure *handwave* 14:56
rudi_s My point is, we shouldn't say it returns a Seq in the docs if it doesn't.
moritz +1
psch and the hyper is what builds the List afaik
14:57 BenGoldberg joined
psch yeah, the nodemap helper for METAOP_HYPER_POSTFIX builds a List directly 14:58
tbh, i'd rather say the example is misleading
the .kv call doesn't happen on a Hash in the first place
rudi_s Yeah.
14:59 apiw joined 15:01 Ben_Goldberg joined, vendethiel left 15:02 Relsak left
psch well, the example is useful on it's own, it just seems a little misplaced vOv 15:02
rudi_s Proposed patch: pbot.rmdir.de/3tRjAtzxfSdFKYbusG166Q 15:03
15:04 BenGoldberg left 15:05 dududuedu left 15:06 Ben_Goldberg is now known as BenGoldberg, apiw left
psch rudi_s: looks good to me, fwiw 15:06
15:07 kurahaupo joined 15:15 kurahaupo left, kurahaupo joined 15:16 apiw joined, kurahaupo left, kurahaupo_ joined
dalek c: d70a800 | (Simon Ruderich)++ | doc/Type/Hash.pod:
Hash: clarify the looping over sorted keys example
 15:18
rudi_s psch: Thanks.
15:19 vendethiel joined 15:21 jack_rabbit joined, Ben_Goldberg joined 15:22 apiw left, Ben_Goldberg left 15:23 BenGoldberg left
jnthn lizmat: On the state/closure thing, agree with moritz; the closure is taken and stored at state variable initialization. 15:23
15:24 Ben_Goldberg joined, apiw joined 15:27 apiw left 15:28 zakharyas joined, apiw joined 15:30 Relsak joined 15:31 _mg_ joined
nine I just don't get it. How can my $custom-lib := nqp::hash(); in CompUnit::RepositoryRegistry ever become empty after having actual contents? 15:32
15:32 zpmorganASDF joined 15:34 zpmorgan left, zpmorgag joined
nine Aaah...different processes! 15:36
15:36 apiw left
nine And of course, the precomp process skips the normal repository initialization, so $custom-lib stays empty 15:37
renormalist is there something I can install to have moer Emacs'ish keybindings in perl6 REPL (having rakudo 2016.01)?
(wordwise jump, delete, transpose typos) 15:38
15:38 apiw joined, zpmorganASDF left
psch apparently i have to build a ContextRef manually to get jvm interop working with all the new CUR stuff..? 15:39
15:39 jack_rabbit left
psch renormalist: i think Linenoise does a few of those at least? otherwise people use rlwrap i think 15:39
the Linenoise module that is*
renormalist looks up rlwrap
15:40 musiKk joined, apiw left
psch re the ContextRef thingy: previously RakudoJavaInterop.computeInterop() returned a Hash, which apparently was enough to register the class successfully. nowadays i run into a nqp::ctxlexpad invocation that wants a ContextRef there 15:41
unfortunately i don't have one, and the existing CallFrame i could stuff into a freshly allocated one apparently don't have the methods actual lexical information about e.g. the methods of the class 15:42
renormalist psch: thanks, rlwrap works for me.
psch renormalist: you're welcome :) 15:44
renormalist and I also realized it is an FAQ - I'm now just reading that whole thing. :-)
psch hm, i might not actually need anything though, right? as in, a java class never has an attached lexical scope, only methods and attributes... not sure if that actually means i don't need CU::Handle.globalish-package to know what a java class can do...
orrr i forgot what exactly a lexical scope is :S 15:45
15:46 apiw joined
nine Yes! "in block at /home/nine/rakudo/install/share/perl6/site/sources/C59503AFF985B18BEF96604D1979FE6EA8004641 (Panda::Installer) line 61" while the precomp file only contains "site#sources/C59503AFF985B18BEF96604D1979FE6EA8004641 (Panda::Installer)" 15:48
RabidGravy nine++ 15:49
15:50 apiw left 15:57 apiw joined
dalek p/relocateable-precomp: 571323a | (Stefan Seifert)++ | src/ (2 files):
Option for a source-name different from the actual source file

While absolute paths are useful for reading source files reliably, we don't want to save those absolute paths in precompiled files as the source files may move after precompilation. This happens when modules are packaged for Linux distributions and the build process runs with an unprivileged user while the result should be installed in a system location.
The source-name option allows for specifying a different string to use for the $?FILES variable and thus for the path to the source file in backtraces.		 15:59
15:59 khw joined
Skarsnik hoo, is that a step to system wide pre comp file? 16:00
stmuk_ nine++ 16:01
dalek rakudo/relocateable-precomp: 59ed434 | (Stefan Seifert)++ | src/core/CompUnit/Repository/Installation.pm: 16:06
rakudo/relocateable-precomp: Turn short-name lookup files into directories
rakudo/relocateable-precomp:
rakudo/relocateable-precomp: This may become part of CompUnit::Repository::Installation format v1.
rakudo/relocateable-precomp: Having to change any already existing files on installation of a module makes
16:06 dalek left 16:07 dalek joined, ChanServ sets mode: +v dalek
nine Store logical file names in precomp files 16:10
We now use logical file names for annotating sources and construct absolute paths when printing stack traces. This avoids storing absolute paths inprecompiled files which is a problem for packaging modules for Linux distributions.
It also allows for us to add the name of the requested module to the backtrace output, alleviating the problem with the undeciferable SHA-1 file names.
Review: github.com/rakudo/rakudo/commit/37...a920730d01
16:11 yurivish left
rudi_s I have either a Str or Nil and want to call .chomp on Str but not on Nil (which causes a warning). I tried $x.?chomp but that doesn't work as Nil is a subclass of Class which has .chomp as method. Any idea? 16:12
16:18 cpage_ left 16:19 apiw left
timotimo m: say Nil.chomp 16:24
camelia rakudo-moar 7010f3: OUTPUT«Use of Nil in string context in block <unit> at /tmp/MbONCtpJkV line 1␤␤»
timotimo ah, that's how
m: say quietly Nil.chomp
camelia rakudo-moar 7010f3: OUTPUT«␤»
timotimo m: say quietly "oh my gosh ".chomp
camelia rakudo-moar 7010f3: OUTPUT«oh my gosh ␤»
16:24 virtualsue joined
rudi_s That feels a big ugly. 16:25
Is there something like $_ = $_.chomp with $x which modifies $x in short? 16:27
timotimo sure, $x .= chomp
oh, that "with" was perl6-with 16:28
rudi_s Ah, nice.
$x .= chomp with $x; good enough for me.
timotimo aye
rudi_s Thakns.
16:28 apiw joined 16:30 domidumont left
timotimo nine: i merged that commit into nom; that was a mistake, wasn't it? :) 16:32
nine: because now make install complains with "Constraint type check failed for parameter '$precomp-id'" "in block <unit> at tools/build/install-core-dist.pl line 13"
nine ouch 16:33
You'd probably need the whole relocaeable-precomp branch 16:34
16:34 _mg_ left
timotimo oh, did i say "merged that commit"? i meant "merged that branch" 16:35
rather than cherry-pick it
maybe i need a reconfigure?
nine You also need the relocateable-precomp nqp branch 16:36
16:37 Ben_Goldberg left 16:39 Ben_Goldberg joined
timotimo oh! 16:40
yeah, turns out i didn't look closely enough at all
rudi_s Can I prefix an enum with a "namespace"? enum Foo <A B C>; adds A directly into the file. I'd like to get Foo::A. At the moment I have module Foo { enum Foo <A B C>; } but then the type is Foo::Foo. Can I get just Foo? 16:42
timotimo >:(
Configure.pl isn't good at NQP_REVISION and branches and merges and stuff
with a straight-up checkout of the relocatable-precomp branch it works better 16:44
16:44 Ben_Goldberg left 16:45 skids joined 16:46 Ben_Goldberg joined 16:47 pierrot joined, ELBeavers_away left
rudi_s Can I use the return type constrain to declare that the function should return an array with values of specific type? Like my Str @x; 16:48
16:48 apiw left 16:50 Ben_Goldberg left
timotimo no, same as parameter type constraints 16:50
16:51 Ben_Goldberg joined
Ben_Goldberg m: my %vowels = 'a' => 1, 'e' => 2, 'i' => 3, 'o' => 4, 'u' => 5; for %vowels.sort(*.key)>>.kv -> [$vowel, $index] { "$vowel: $index".say } 16:52
camelia rakudo-moar 7010f3: OUTPUT«a: 1␤e: 2␤i: 3␤o: 4␤u: 5␤»
rudi_s timotimo: But I can declare foo(Str @foo) and it seems to work.
I want the same for the return type. 16:53
timotimo that is a check for "does the user pass a Str-typed Positional of some kind"
it does not check "is the array passed in filled with only Str objects or derived from Str objects"
so if you do "returns Positional[Str]", you'll have to declare the object you're returning to fill that role
it won't introspect a regular array for "is it only Str in there?" 16:54
rudi_s Ah, too bad. Thanks.
Ben_Goldberg m: my %vowels = 'a' => 1, 'e' => 2, 'i' => 3, 'o' => 4, 'u' => 5; for %vowels.sort(*.key) -> (:key($vowel), :value($index)) { "$vowel: $index".say } 16:55
camelia rakudo-moar 7010f3: OUTPUT«a: 1␤e: 2␤i: 3␤o: 4␤u: 5␤»
16:58 apiw joined
timotimo nine: i wonder if we can get the profiler to know about the nicer filenames, too 17:00
nine: but the profiler itself is "only" vm-level, so not exactly sure how to introduce the two 17:01
17:03 jameslenz left
timotimo now that our filenames are so super long, maybe it'd be a good idea to have the filenames as a list up front in the json blob and then reference that array later on 17:03
so that the file gets a bit smaller?
17:04 jameslenz joined
timotimo on the other hand, i really can't tell what exactly it is that makes browsers fall to their knees: raw json blob string size, or size of the constructed object? 17:06
Skarsnik the js is bad maybe§
timotimo if you'd like to improve it, feel absolutely free :) 17:07
Skarsnik I rather work on the Qt thingy x)
17:07 apiw left 17:08 jameslenz left, musiKk left
timotimo sure, that's fine, too 17:08
last time i looked, it didn't do anything with the GC list, for example
and it doesn't do anything at all with the call graph except sum up all the data 17:09
17:09 jameslenz joined
flussence I wonder how many of these workarounds are going to pile up before we all finally accept the sha1 filenames thing is insane :/ 17:10
timotimo i'd find it nicer if the filenames could be the original filenames on disk and only change if there's disambiguation needed
but there could definitely be good reasons against my idea 17:11
MadcapJake . 17:13
yoleaux 13:30Z <pmurias> MadcapJake: do we need %Test::Lab::context for anything? in ruby the defaults are set per class, having that global doesn't seem to give us much
17:14 apiw joined
MadcapJake .tell pmurias the Ruby version has the helper method as a module, so the default_scientist_context lets you change the context for any time you use the science helper. I thought the best reflection of that in P6 would be a module-scoped hash. 17:16
yoleaux MadcapJake: I'll pass your message to pmurias.
ugexe timotimo: i imagine part of it might be you don't really know whats allowed (at least on windows). on windows you can create files with illegal characters using perl (5 or 6) and windows does nothing to stop it or warn about it
Ben_Goldberg Err, if a file is creatable with some specific character as part of the name, doesn't that mean that that char is, in fact, legal? 17:17
MadcapJake .ask pmurias what would you see as a more Perl 6 reflection of that? Check here for some context (pun intended): github.com/github/scientist#adding-context 17:18
yoleaux MadcapJake: I'll pass your message to pmurias.
17:19 apiw left
timotimo ugexe: okay, so we'd just limit to a-zA-Z - and _ 17:19
ugexe its legal for some underlying part of windows, but not others
timotimo and - and _ are only allowed between a-z and A-Z
and 0-9
ugexe but without a unique mapping couldnt 2 modules with names differing only by 1 Unicode character end up in the same path? 17:20
timotimo yeah, so? in taht case we disambiguate with whatever method we feel necessary
jnthn I wonder how many discussions we're going to have trying to work out clever mapping strategies before folks learn to darn cope with the sha-1 names. :/ 17:21
MadcapJake jnthn, do you really think sha-1 names in an error message are user-friendly?
17:21 verzit joined
jnthn MadcapJake: No, which is why I said earlier I thought we should show the source filename as it was in the installed distribution. 17:22
ugexe 11:10:33 nine | It also allows for us to add the name of the requested module to the backtrace output,
timotimo yeah, we do that now
jnthn OK, good. 17:23
17:23 apiw joined
timotimo i see no good way to introduce it to the profiler 17:23
jnthn We might want to abbreviate the SHA-1s, or only show them in debugging output.
timotimo often enough the table in the "routines" tab is all wonky because a few routines in the table are super super wide because of their file name
MadcapJake jnthn, ok I'm in agreement with that, i think we should show not only the source but the module-relative path e.g., «Module::Name/Foo/Bar.pm6» 17:24
timotimo and the allocations popup is *super* b0rked, too
jnthn MadcapJake: Yes, that's what I mean by the source filename as it was, before we installed it. Though I can see I that wasn't so clear how I wrote it ;)
mst needs to get round to documenting the current CURLI on-disk layout 17:26
... and then take a chainsaw to it, because having the filenames be hashes is frigging ridiculous
jnthn *sigh*
mst (having a hash as part of the *path*, less so)
jnthn So long as I don't have to work on this, I guess it's fine.
mst jnthn: it's ok. I understand why it was done that way. the trick will be to keep all the advantages of that *and* get back some of the things we lost 17:27
and I'm about 98% sure I know how to do that
I'm also not expecting anybody else to
timotimo how terrible would it be to reduce the shasums to only 2/3rds or 1/2 length? 17:28
17:28 domidumont joined, kid51 left, verzit left
jnthn I guess we just have different design ideas on it. I like the SHA-1s in the installation repo because (a) filesystem valid name issues go away and (b) it makes it darn clear you're not meant to be messing with those things by hand 17:28
MadcapJake jnthn, what's the route to take if you want to tweak a module? 17:29
mst if you're trying to tweak then you want a lib/ dir or something outside the CURLI with the code in it
17:30 perturbation joined
mst similarly to in perl5, where I do 'mkdir lib/Module; cp $(perldoc -l Module::Name) lib/Module/Name.pm; vi $_' 17:30
ugexe just sha1 non-alpha numerics! 17:31
mst jnthn: those are good reasons to have the sha1 as *part* of the path, however in terms of various sorts of debugging having that be the only thing encoded in the path is flipping terrible
basically, (b) is not a reason to be actively hostile to people trying to debug dependency issues 17:32
jnthn MadcapJake: Tweaking installed things sounds like a really bad idea...
MadcapJake it'd be nice if there was a `panda develop Module::Name` that would download (or move from precomp) the module into a directory at the current location and symlink it into the precomp module dirs.
mst jnthn: yeah, it's an absolutely terrible idea at the moment. and frankly I'd rather make sure there are good alternatives so you don't ever want to do that, than make it less of a terrible idea :) 17:33
MadcapJake jnthn, i guess i'm coming from Node.js world where every package is placed in a node_modules dir and you can tweak to your hearts content, I really wish Perl 6 had some similar facility (tweaks on a per-project basis)
flussence what's the issue with just percent-encoding the non-alnum parts of what's currently hashed...?
mst oh gods please no let's not bikeshed this now
we want the hash *as well*
there's lots of useful reasons for this 17:34
the current design is academically elegant and has lots of useful properties
I just need to beat it with a shovel until it's also not actively hostile to sysadmins and operations staff
jnthn mst: To me having a unique identifier for the thing being loaded, so I can know exactly which one it is, is highly useful for dependency debugging. :-)
mst jnthn: WHICH IS WHY I WANT TO KEEP THE HASH AS WELL!
jnthn mst: Yeah, I know :) 17:35
mst sorry. every time I talk about this somebody assumes I want to get rid of the hashes
and, no, they're also useful
jnthn If you were trying to get rid of them I wouldn't be saying "I don't want to do it", I'd be saying "please don't do it". ;)
flussence erm, isn't the *input* to sha1 already a unique ID? that's what I'm saying
17:36 hacst left
mst the last time I designed a system like this, what you basically got was DistributionName-1.23-sha1goeshere/File/Name.pm 17:36
MadcapJake so here's a common workflow that I have in the JS world, install a package, write up my code that uses it, notice something that needs fixing, fix it in my project's node_modules dir, run tests, test my project, then submit those fixes as a PR.
mst which gave you the unique identifier stuff nicely, while still also giving you sensible names on disk to find stuff using 17:37
17:37 hacst joined
jnthn Anyway, my (apparently wrong) assumption was storing the things under hashed names then buliding tooling around that would be a fine the way to go. But if having the stuff on disk more closely match the original file names is really valuable, I can go with us adding that too. 17:37
17:38 apiw left
jnthn cooking needs attention; bbiab 17:38
mst jnthn: there's lots of stuff that relies on file names already out there, and lots of common debugging tricks that use them 17:39
as an example, if I get an error of "can't find Foo::Bar" or so
for perl5, I can find out if that's in a vendor package by running 'apt-file search Foo/Bar.pm'
17:39 alpha123 joined, pierrot left
mst it would seem quite sad to say "yeah, you can't do that for perl6, you'll have to make your sysadmin learn a complete new set of ways of working" 17:40
17:40 sjoshi joined
mst because what I'm willing to bet will happen in a bunch of cases is the response you'll get it "if you're going to be this hostile to basic systems stuff, I'm not deploying this in production because I don't want to deal with this if I get a 3am page" 17:40
and, honestly, I'm pretty sympathetic to that response
so I want to make it unnecessary, without losing any of the cool advantages of the current system :) 17:41
timotimo i'm glad we have mst working for us :3
mst gets off his soap box
mst still isn't expecting anybody else to do the legwork
flussence what timo said +1
mst if nothing else because if somebody else does it I'll probably only moan about their version 17:42
first weekend I don't have a frigging head cold, I swear I'll look at this properly
meanwhile, you're welcome to blame me for it not being done already :D
flussence (if anything I've said sounds idiotic, feel free to ignore. bit under the weather myself today) 17:43
mst percent encoding is a bit icky when you're trying to navigate around the filesystem 17:44
timotimo ugh, yes
mst we'll probably have to experiment a bit with what to do instead of that, it's not a terrible plan Z 17:45
a completely random thought would be: if you have something like Cote.pm where there's a circumflex over the o
you'd encode it to Cote-abcdef.pm where abcdef is the first 6 chars of the sha1, with code that'll use more characters if it somehow manages to collide within its target directory 17:46
note: just thought of that, idea not actually endorsed, meant as a thought to consider
but something in that direction would let us avoid the filesystem-character-restriction problem while still having reasonably ops-friendly names
(goal: let somebody, when paged at 3am, manage to debug the broken production service even if they're not awake enough to remember perl6 has a whole extra set of tools for interacting with this stuff) 17:47
MadcapJake: we need to enable some approach to hacking stuff up, but I'd rather think about it because once we make it even possible, people will do it all the time and it's effectively supported forever whether we meant that to happen or not :) 17:48
17:54 virtualsue left
MadcapJake mst, yeah that's a good point! Better to have a well-engineered path that can stand the test of time. 17:57
18:00 sjoshi left, apiw joined 18:01 sufrostico joined, sjoshi joined
sufrostico hi there 18:02
somebody knows if there is a way to incorporate C code whitin a perl6 module ?
other than the NativeCall library ?
mst "is there a way to do X, other than the way to do X?" 18:03
why do you need another way? :)
DrForr Why won't NativeCall work for you?
18:03 apiw left
simcop2387 w 60 18:03
18:05 hacst left
sufrostico there is some code that not exactly compiles to a shared library (.so) 18:06
18:07 hacst joined
sufrostico the code is not mine and the manual basically says download the code to your project and add the files to your makefile 18:07
so, i was checking the options
18:08 adu joined 18:10 TEttinger left, vendethiel- joined, kst`` joined, wamba left 18:11 virtualsue joined
timotimo there's Inline::C with which you can have a sub that has C source code as its body 18:11
18:11 d4l3k_ joined, dalek left, vendethiel left, Gothmog_ left, Roamer` left, d4l3k_ is now known as dalek, ChanServ sets mode: +v dalek, Actualeyes left 18:12 wamba joined, kst` left, ilbot3 left, MilkmanDan left, apiw joined, Gothmog joined 18:13 Gothmog is now known as Gothmog_, Roamer` joined, zakharyas left
mst sufrostico: you could still build them into a .so 18:14
sufrostico: and you're going to need a .so to be able to runtime load them into perl6 no matter how you do it, I think 18:15
sufrostico: though Inline::C may be able to do that for you
18:16 apiw left
sufrostico mst: Is posible that the perl6 module compile the c sources when installing ? 18:16
ugexe put code to compile the source in a Build.pm 18:17
18:18 apiw joined
sufrostico ugexe: Thanks, I'll look into that 18:22
timotimo we have things in the ecosystem that help you with that, like LibraryMake 18:23
18:24 TEttinger joined 18:26 apiw left
RabidGravy yeah I have three modules that do it, it's fairly easy to do 18:27
(build a small C library that is)
18:28 wamba left
ribasushi putting this out here again (perl.christmas) twitter.com/ribasushi/status/701468980855382016 18:29
timotimo now if they had a 6 in there ... :P 18:32
sufrostico RabidGravy: you have the urls of the modules at hand ? 18:33
I'll love to take a look at those 18:34
18:34 adu left
timotimo why not just modules.perl6.org ? 18:34
oh, sorry
RabidGravy github.com/jonathanstowe/p6-Sys-Utmp, github.com/jonathanstowe/p6-Sys-Lastlog, github.com/jonathanstowe/RPi-Device-SMBus for example
timotimo you didn't mean for LibraryMake, obviously 18:35
18:35 apiw joined
RabidGravy see also Inline::Perl5 that does similar 18:35
DrForr And Inline::Scheme::Guile. 18:36
ugexe you can remove about 5 dependencies from your overall dependency chain by using a Build.pm modeled this way: github.com/retupmoca/P6-Compress-Z...r/Build.pm 18:37
18:40 apiw left 18:41 ilbot3 joined
jnthn mst: Thanks for the explanations; they make sense. 18:42
mst jnthn: ISTR you spend quite a bit of time in win32? 18:43
or am I wrong about that? 18:44
timotimo i think that's fair to say
mst I think if you don't live in unix full time you don't see the extent to which directory/path naming when done carefully is actually a massive part of the utility 18:45
jnthn mst: Yeah, quite a bit.
mst for you, needing special tools likely isn't a surprise or a problem so long as they're good
18:45 firstdayonthejob left
jnthn mst: And while I do increasingly more of my day to day development work on Linux, it's fair to say I've never had to admin anything much more than a server hosting a handful of simple websites. :) 18:46
mst for me, it would be a moment not unlike sungo.im/misc/fault-tolerance.png (swearing)
18:46 Relsak left
jnthn mst: Though curiously, my design inspiration for it was actually largely Git 18:46
mst: Which is just a big store of things addressable as SHA-1s. :) 18:47
mst if git used the shas in the *checkout* I'd want to murder it too
jnthn Yeah :)
Context, context, context... :)
mst oh, yeah, also - a very trivial example - grep gives you the file name and the matching line
grep -r 'use Foo' $CURLI_PATH
flussence (git also has, according to my bash-completion, 170 frontend programs to that on disk format) 18:48
mst would produce rather useful output under the approach I'm thinking about
whereas currently you're a bit stuffed
jnthn *nod* 18:49
Yeah, I see why you'd miss it.
mst and, like, it's nice to have shiny thing-specific tools but if production just blew up I do not want to -have- to remember them
sufrostico timotimo, mst , ugexe : thanks 18:50
mst jnthn: excellent. if your POV is now "this isn't my itch, but I could understand why somebody else would want to scratch it, and that if it goes well the end result would be a net usability improvement" then I'm happy 18:51
jnthn mst: Yes, thanks for being patient enough to fill me in on what I was missing.
18:52 john51 joined 18:53 john51 left 18:54 john51 joined
mst jnthn: thanks for being part of creating something with a developer experience nice enough for me to care :D 18:55
flussence
.oO( oh this is neat, some of the git frontends are hardlinked by category: `ls -l /usr/libexec/git-core | sort -nk 2` groups them by functionality [loose ends, servers, net transports, user interface] )
mst cute!
stmuk_ I don't think anyone should cite git as an example of well designed front ends however neat the backends are :) 18:56
see git-man-page-generator.lokaltog.net/ 18:57
mst well, I think it's a good example of "how to make something that's extremely powerful once you know it backwards, but the learning curve is steep and you need to keep it memorised" 18:58
jnthn: which is kinda what you did, and the one part of git I don't want us to accidentally steal :D
18:59 john51 left
MadcapJake .tell perlpilot I got another email about GSoC from Mark Keating, they decided to hold off this year and start preparations in the fall to make for a stronger, more-organized effort for the 2017 GSoC 18:59
yoleaux MadcapJake: I'll pass your message to perlpilot.
19:00 john51 joined
jnthn stmuk_: Agree, though a good backend can have another frontend put on, but a bad backend design is probably pretty hard to redeem with a better frontend. :) 19:01
mst MadcapJake: that sounds like an mdk sort of plan 19:03
19:03 virtualsue left
MadcapJake mst, I don't know mdk but I do think that the redoubling should've happened 2015 Fall not 2016 Fall. 19:05
mst MadcapJake: remember I explained that the summer 2015 picks were basically inexplicable and put a lot of people off - we didn't have the available volunteer time+motivation to do that as a result 19:06
I do think it would've been nice if we could've done it in fall 2015, but you can't force people to be motivated, especially after they've just been shat on from a great height 19:07
awwaiid I don't have time myself... but just to throw it out there -- we could have our own (non-google-sponsored) internship type program if we could scrape together the time and $$
19:07 kid51 joined
awwaiid though maybe google puts a lot more $$ in this than I'm realizing 19:07
labor-wise, they have the orgs put up details, find students, mentor students, evaluate students ... 19:08
19:08 MilkmanDan joined
mst hm. what might be interesting is to submit for 2017 ... and then if we get refused, open it up to companies to pledge to sponsor the students anyway 19:09
awwaiid exactly
says here it is $5500 per student that they get
mst MadcapJake: eventually, you discover that 'should' generally isn't a useful word for such things
MadcapJake mst, but skipping just seems kind of immature 19:10
mst because it's effectively complaining that other people didn't volunteer to do the things you wanted them to do
awwaiid (which... is about the size of the budget for the DC-Baltimore Perl Workshop... so maybe nontrivial)
MadcapJake i'm speaking entirely from the perspective of what I would expect a Foundation to be
stmuk_ I don't think Google give the same orgs money every year anyway .. they try and spread it around 19:11
mst it's still made of volunteers, and better to do it next year properly than to half ass it this year and waste all the time put in
MadcapJake mst, certainly a valid perspective, I'm not disagreeing, I just thinking from the perspective of those at Google deciding these things, for a well-recognized foundation to skip a year seems a bit "non-foundation-like" 19:12
mst people seem to have expectations of TPF that really don't match the reality of it being a non-profit run by volunteers in their spare time (and note also that on the occasions we've proposed fixing that by paying people, a whole different set of people have got up in arms about it)
and, as I said earlier, after the debacle last year quite a number of projects are skipping a year 19:13
I suspect some of them won't bother again full stop, if the process doesn't start making a bit more sense
MadcapJake mst, perhaps that's there intent :P 19:14
jnthn Don't ascribe malice to what could easily be disorganization :)
mst perhaps. but I don't expect anybody who's actually worked with OSS non-profits to be remotely surprised, either way
I think it's just that they completely changed the selection process, and forgot to tell anybody what they were doing or how 19:15
I mean, it's google. it's not like they haven't done that before.
jnthn :-)
mst MadcapJake: if it was six months ago and you were volunteering to help co-ordinate, this would be a different conversation - but as it is, all you're really achieving is bagging on volunteers for finding that google's behaviour killed their -Ofun for a bit 19:16
19:19 AlexDaniel joined
MadcapJake mst, not trying to bag on volunteers! Sorry! I wish I was involved six months ago to help! 19:19
19:19 john51 left 19:20 john51 joined
mst MadcapJake: the trick is to remember that TPF itself is a legal vehicle which exists to hold and distribute money - and then everything else done under its auspices is done by volunteers 19:21
at which point, you end up with more accurate expectations of what's possible :) 19:22
MadcapJake mst, sure, however I do think that part of the responsibility of existing as a legal vehicle is recognizing that you are (uncontrollably) going to be a cultural vehicle for Perl efforts and others will look to you for the current health and vitality of Perl 5/6. 19:26
mst I'm sure the people involved are very much aware of that, but sadly knowing this fact does not magic volunteer time and motivation out of thin air 19:28
MadcapJake unfortunately ;)
19:30 sjoshi left 19:34 john51 left, _dolmen_ joined, john51 joined 19:38 gtbjj joined 19:40 yqt joined 19:42 Ben_Goldberg left 19:43 Ben_Goldberg joined 19:48 Ben_Goldberg is now known as BenGoldberg 19:51 kid51 left 20:00 Cabanossi left
pmurias MadcapJake: re having a global %context I don't have a good proposal for a replacement, it just doesn't seem very useful and seems to be something that's not strictly necessary 20:01
yoleaux 17:16Z <MadcapJake> pmurias: the Ruby version has the helper method as a module, so the default_scientist_context lets you change the context for any time you use the science helper. I thought the best reflection of that in P6 would be a module-scoped hash.
17:18Z <MadcapJake> pmurias: what would you see as a more Perl 6 reflection of that? Check here for some context (pun intended): github.com/github/scientist#adding-context
MadcapJake pmurias, but what do you find not useful about it? 20:02
20:02 donaldh left
pmurias MadcapJake: in ruby you can have a set of per class defaults 20:03
if I wanted to pass context stuff in globals instead of using a global %context hash I would just use a global 20:04
20:04 Cabanossi joined
MadcapJake but the %Test::Lab::Context isn't global it's module-scoped 20:04
maybe another route would be to add a named argument `context` 20:05
20:05 donaldh joined
pmurias you mean to the dsl-free &lab version 20:05
MadcapJake right 20:06
pmurias for the dsl using lab version you can just send things using $experiment.context
just like in ruby
MadcapJake: I plan on adding a context argument
20:07 telex left
pmurias what I mean about %Test::Labl::context being useless is that in ruby it's per class, so in CuteWidget you can add a bunch of context that's shared for all experiments on cute widgets 20:08
20:08 telex joined
MadcapJake oh i see, you mean that in ruby the science helper is added by mixing a module in 20:08
and that module has the context method that you can change 20:09
20:09 _dolmen_ left, domidumont left
pmurias the class you are mixing things in can have a default_scientist_context 20:09
MadcapJake but in my Test::Lab, it's module-scoped but really, since my module isn't a role, it's just the same as being globally scoped. is that what you mean?
pmurias yes
if we had a role that we added it would make sense to also have a default_lab_context 20:10
MadcapJake ok i'm following, but i'm kind of against it being a role because unlike ruby, you'd have to call it as `self.lab` 20:11
pmurias role Test::Lab::UseLabAsMethod {method lab(*%args) {lab(:context(self.default_lab_context), |%args}; method default_lab_context {...}} 20:12
maybe with a few tweaks and a better name ;) 20:13
MadcapJake so add that as an alternative?
20:14 cdg joined, average joined
average blog.garage-coding.com/2016/02/05/b...queue.html 20:14
anyone here want to skim through my blog post 20:15
pmurias I think it would be best to first simplify things and then maybe add frills latter
20:15 cdg left
pmurias I have started working on a dsl-free lab, but wasn't sure how to deal with context 20:15
MadcapJake pmurias, what's your current thinking on context then? 20:17
20:18 von_cheam joined, cdg joined
MadcapJake also, was thinking another route to take for `try`, instead of having a named `:name` argument, we could allow the user to pass a named sub `.try: sub candidate1 { ... }` 20:19
20:20 _mg_ joined 20:21 cdg left
von_cheam Hello! Complete newbie question here. Is it possible to create a static-type array in Perl 6? I can't find anything online.. 20:21
20:21 cdg joined
Skarsnik static type? you mean an umutable Array? 20:22
20:22 darutoko left
MadcapJake von_cheam, what do you mean by static-type? 20:22
Skarsnik List or Seq are unmutable
MadcapJake Skarsnik, do you mean immutable?
Skarsnik yes probably
von_cheam As in something like: 20:23
my Int @value;
That'll only accept Integers as elements.
MadcapJake that's exactly how you do it von_cheam
hippie1 m: my Int @foo; @foo.push("steve")
camelia rakudo-moar 7010f3: OUTPUT«Type check failed in assignment to @foo; expected Int but got Str ("steve")␤ in block <unit> at /tmp/_cGxzNDFjs line 1␤␤»
von_cheam Oh, okay, fantastic. Thanks very much! 20:24
hippie1 Perl 6 is so good, you know it without even knowing you know it.
von_cheam Haha! 20:25
20:25 von_cheam left 20:26 ely-se joined
MadcapJake hippie1, that's going on twitter :D 20:27
twitter.com/perlhex/status/701503672547569664 20:28
Begi We have to wake up the #perl6 hashtag !
20:28 rindolf left
MadcapJake Begi, there's a light use of it already, but surely not enough! 20:28
20:28 sufrostico left
Begi that's sure, not enough ! 20:29
20:29 john51 left 20:30 john51 joined, sufrostico joined
shadowpaste "pmurias" at 217.168.150.38 pasted "how a DSL free &lab would work" (7 lines) at fpaste.scsys.co.uk/505999 20:31
pmurias MadcapJake: ^^ how &lab works in my working copy 20:32
MadcapJake: instead of having a bunch of "cute" pseudo accessor we could have &lab for the suger and make Test::Lab::Experiment a normal object
20:33 rindolf joined
MadcapJake pmurias, i like that example, but there's no `use`, is that intentional? 20:33
pmurias, I think that's a great idea, do you have an example of what that would mean in practice? 20:34
pmurias I'll paste a version of use
s/of/with/
shadowpaste "pmurias" at 217.168.150.38 pasted "how a DSL free &lab would work" (23 lines) at fpaste.scsys.co.uk/506000 20:38
pmurias MadcapJake: ^^ a more complex example
MadcapJake part of me is still a bit hesitant to turn lab into a named-arg sub as the ruby version allows you to do anything you want inside the procedure block, while the named-arg perl6 routine would only allow you modify thing internal to each named arg or before called &lab.
20:39 Begi left
MadcapJake what are your feelings on this pmurias? 20:39
pmurias can we do anything inside the procedure block other than set the named arguments? 20:40
20:40 Begi joined
MadcapJake pmurias, currently you could do anything you want but really only for preparation as none of the tests are run until after the procedure finishes. 20:41
pmurias you will be still be free to calculate the named arguments how you want outside the lab block 20:42
MadcapJake but maybe you wanted to dynamically generate something, then you'd have to add that outside of the lab or do it inside each use/try/run-if etc.
yeah true, the facility is still there but i can't decide what's a better interface
pmurias any use cases when you want to dynamically generate stuff inside the lab? 20:43
MadcapJake the benefit of the DSL are (to me): 1) simple wrapping 2) less typing 3) a block to run anything you want when declaring your experiments 20:44
the benefits of named-args are: 1) more perlish 2) more readable 3) less indenting (and in some cases less typing) 20:46
pmurias what do you mean for simple wrapping? 20:47
MadcapJake you only have to place one line before and a curly end after. whereas with a named-args design you would need to rewrite the code inside of a call to &lab 20:48
pmurias don't you have to just type less to wrap? lab 'test', { .use: {...}} vs lab :name<test>, use => {...} 20:52
b2gills (backlogging) Why does everyone use ｢%hash.sort(*.key)｣ instead of just ｢%hash.sort｣?
pmurias and we could also use lab 'test', use => {...} in the DSL-free version
20:52 _mg_ left
Skarsnik sort call %hash.flat.sort ? 20:52
pmurias so it would be lab 'test', { .use: {...}} vs lab 'test', use => {...} 20:54
MadcapJake yeah that certainly is pretty close :P 20:56
20:58 john51 left, john51 joined
lizmat fg 20:59
20:59 musiKk joined
lizmat oops :-) 20:59
spider-mario I see that we have the same kind of problem :) 21:00
b2gills I don't have that problem because my terminal is white on black, and Pidgin is by default black on white
spider-mario I occasionally type `ls` into my hangouts window with my “main” coworker
promptly followed by “oops, sorry” 21:01
to which she replies “np”
moritz :w 21:02
no vim bot around :-)
spider-mario :x
21:08 geraud joined 21:09 dh7320 joined
nine timotimo: I guess the profiler just uses the <file> annotation of the MAST nodes? Those do now contain the requested module's name in parentheses in my branch. Should be somewhat trivial to extract that, shouldn't it? 21:17
21:18 Skarsnik left 21:19 Amendil left 21:20 ely-se left
timotimo oh. yeah, it uses that 21:22
21:23 kaare_ left 21:38 firstdayonthejob joined
BenGoldberg b2gills, %hash.sort(*.key) merely happens to be what's in Hash.pod. 21:40
m: my %vowels = 'a' => 1, 'e' => 2, 'i' => 3, 'o' => 4, 'u' => 5; for %vowels.sort -> (:key($vowel), :value($index)) { "$vowel: $index".say }
camelia rakudo-moar 7010f3: OUTPUT«a: 1␤e: 2␤i: 3␤o: 4␤u: 5␤»
21:42 Ben_Goldberg joined 21:44 ely-se joined, rpburkholder joined
Begi I'm a beginner with Perl 6 : is this code correct ? 21:45
pastebin.com/GPzwzu1y
21:45 BenGoldberg left
Begi Can I improve it ? Thanks ! 21:45
MadcapJake prepare doesn't need to be a multi if there's only one sub 21:46
personally, i would just inline that if you're not planning on using `prepare` anywhere else 21:47
21:48 TEttinger left, rburkholder left
Begi Oh yes, why not. 21:48
Hmm, I'm gonna use it ine the future in fact
MadcapJake also you don't need the `$_ eq` as `when` automatically checks a string for `eq` against the topic variable 21:49
21:50 TEttinger joined
Begi so, I can do something like when 'dec' {} ? 21:50
MadcapJake instead of `say 'Invalid!'` maybe you want to use `die 'Invalid'` which will exit with an error code and a stack trace
21:50 SCHAAP137 joined
MadcapJake `when 'dec' {...}` 21:50
Begi nice idea ! and yes, that's it 21:51
thanks !
21:55 Begi left
MadcapJake also, the MAIN doesn't need multi, but you *could* separate the modes and do a different multi for each mode 21:55
e.g., «multi sub MAIN('enc', Int $n, Str $filename) { ... }» 21:56
multi is only needed where you plan on supporting multiple Signatures i.e., different parameters per routine 21:59
22:02 wamba joined 22:03 FROGGS left 22:04 sufrostico left 22:12 john51 left, john51 joined 22:14 gtbjj left 22:15 dh7320 left 22:22 xnrand left, xnrand joined 22:23 john51 left, john51 joined 22:29 john51_ joined 22:31 _dolmen_ joined 22:33 john51 left 22:34 musiKk left
dalek osystem: 8b44800 | (Aurelio Sanabria)++ | META.list:
Add RPI::Wiring::Pi to the ecosystem

See github.com/Sufrostico/perl6-wiringpi/		 22:39
osystem: df996f8 | RabidGravy++ | META.list:
Merge pull request #158 from Sufrostico/master

Add RPI::Wiring::Pi to the ecosystem
RabidGravy There are a goodly few raspberry pi modules getting in there 22:40
this is a good thing
22:43 john51 joined 22:46 john51__ joined, john51_ left
AlexDaniel “(b) it makes it darn clear you're not meant to be messing with those things by hand” – eh, too bad the easiest way to fix a module sometimes is to open the goddamn file and edit it… 22:50
22:51 john51_ joined, john51 left
AlexDaniel it is a bad idea, yeah, whatever. It works 22:51
at least right now 22:52
22:53 rindolf left 22:55 john51__ left 22:59 neilb joined 23:00 pmurias left 23:02 Ben_Goldberg left 23:04 ely-se left, john51 joined
AlexDaniel average: hi! So I skimmed through your blog post 23:05
MadcapJake AlexDaniel, that's why I wish for a «panda develop Module::Name» that would pull a module out of it's sha'd home for you to develop (but symlinked back into the sha'd dir)
AlexDaniel MadcapJake: yeah, that'd be great 23:06
my biggest problem right now is that stacktraces don't make any sense to human beings
so I'm glad that we have mst++
MadcapJake yep that's exactly where the discussion earlier came from
AlexDaniel that being said, when stacktraces are finally fixed I'll have other problems… :) 23:07
average: ok, so, since it is about bash… prepare for bash pitfalls!
23:07 john51 left
MadcapJake honestly my two grievances right now are 1) stacktraces are practically useless and 2) hacking on modules is discouraged and difficult 23:07
23:08 john51 joined, john51_ left
AlexDaniel average: a rule of thumb is to quote everything. E.g. here ｢. $PWD/queue-samples/seq-stats.sh｣ 23:09
average: or here: echo "$MESSAGE" >$FIFO
average: or here: rmdir $CONSUMER_LOCK
average: everywhere. Everything has to be quoted. Although unquoted things may work right now, there's no guarantee that a tiny space will never get in 23:10
MadcapJake what's the best way to run a p6 script in the background at intervals (on linux)?
AlexDaniel average: Okay, now this: ｢read line｣. read without -r does not make much sense 23:11
23:11 john51 left
AlexDaniel average: wiki.bash-hackers.org/commands/buil...without_-r 23:11
23:12 john51 joined
AlexDaniel average: ok, now this is probably not as safe as you might think: echo "$MSG" 23:12
23:13 RabidGravy left
MadcapJake oh nvm, «watchbb 23:13
lol oops, meant to say «watch» works
AlexDaniel average: I have no idea what could possibly be inside "$MSG" but try ｢MSG='-n'｣ and see how it breaks
average: so yeah. In short: 1) always quote everything 2) always use read -r 3) use printf when printing arbitrary variables 23:15
23:16 john51_ joined 23:17 lichtkind joined
AlexDaniel average: I might take another look when you fix these things 23:17
lichtkind masak cheers
23:17 john51_ left 23:18 cdg left, donaldh left, dh7320 joined, john51_ joined
AlexDaniel average: also, this page is extremely useful in such cases: mywiki.wooledge.org/BashPitfalls 23:19
23:20 john51 left
dalek kudo/nom: 77581b1 | lizmat++ | / (3 files):
Make @a.chrs 3x as fast, chrs(@a) 9x as fast
 23:25
timotimo damn. 23:30
lizmat damn? 23:32
timotimo that's good savings 23:34
japhb lizmat++ # Lots of optimization commits
timotimo "damn those commits are good" 23:35
japhb Very nice things to see on return from vacation. :-)
lizmat FWIW, I'll be glad to revert them once we have deeper optimizations making these obsolete
but I came across this one while working in the p5 pack/unpack module 23:36
which I hope to have ready by tomorrow
23:37 pierrot joined
japhb oooh, nice 23:38
23:38 _dolmen_ left
lizmat github.com/lizmat/PackUnpack fwiw 23:40
23:41 kurahaupo_ left
lizmat preliminary set that will be supported: a A C h H I L n N Q S v V x Z 23:41
timotimo oh, was there anything you could steal from FROGGS' v5?
23:42 pierrot left
lizmat I've taken my inspiration from src/core/Buf 23:43
fwiw, I figured FROGGS would be using that in v5, was I wrong ?
timotimo no clue 23:44
i just guessed
lizmat hmmm... seems it is a different implementation altogether 23:46
will use that as further inspiration tomorrow :-)
timotimo++ for reminding me, FROGGS++ for putting in the work so far 23:47
23:47 pierrot joined 23:49 firstdayonthejob left
timotimo cool :) 23:50
23:50 pierrot left
timotimo i'm about to head to bed 23:50
lizmat too 23:51
good night, #perl6!
23:53 sufrostico joined, pierrot joined 23:54 spider-mario left 23:58 pierrot left

Powered by Cro and the Raku® Programming Language.

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

Thank you!