»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
00:01 dalek left 00:03 dalek joined, ChanServ sets mode: +v dalek 00:10 ajr_ left 00:12 a3gis joined 00:13 thou left 00:16 tadzik joined
dalek osystem: 5432189 | (Andrew Egeler)++ | META.list:
Add Net::SMTP
 00:18
Mouq retupmoca++ 00:22
retupmoca Mouq: It's pretty simple right now, but hopefully it will advance quickly 00:23
00:24 colomon joined
retupmoca Of course, what I want now is a module that does SSL or STARTTLS, but I don't know if I'm comfortable taking that on quite yet 00:24
Mouq retupmoca: Heh, I understand that 00:28
Hopefully I'll have tuits eventually
Also, lue, from a few nights ago: gist.github.com/lue/9941658#comment-1204497 00:36
00:40 hoverboard left 00:41 a3gis left 01:00 virtualsue joined 01:01 klapperl_ joined 01:03 klapperl left 01:09 hoverboard joined 01:24 virtualsue left 01:37 xenoterracide left 01:38 thou joined 01:44 ilbot3 left 01:46 ilbot3 joined 01:52 hoverboard left 02:03 klapperl_ left 02:05 klapperl joined 02:09 xenoterracide joined
lue Mouq: saw it. I'm having serious doubts that pack/unpack have any use (its only real strength, interchange with lower APIs, isn't *too* hard to do already). I'd like to see what TimToady's thoughts are on the matter. 02:10
Mouq lue: He's talked about it before…
lue Also, am I the only one getting a X::Undeclared::Symbols when trying to compile perl6-p's setting? 02:11
Mouq: really? haven't seen it then.
I only see two routes for (un)pack: first is to not have it, which might annoy some people who like its conciseness, but that leaves open the chance for methods on Buf to do pack-like things ( Buf.push($a, :uint(32)) ?), or... 02:19
have a whole 'nother sublang, which as I stated earlier feels like overkill. 02:20
geekosaur there is a lot to be said for erlang's bit syntax...
Mouq lue: Note that we only implement a shallow subset of the old pack syntax
lue Mouq: sure, I doubt those BER integers thingys would be very useful :) 02:21
Mouq lue: Well, p5 has sub-tamplates 02:22
*templates
lue (the advantage of some PACK lang is that some could extend it to include those BER integers if needed, whereas a pack() method or sub would need to be overriden to do same, if it goes the P5 route.)
Mouq And has ">" and "<" to indicate byte-order 02:23
So p5 is already doing one-pass parsing
(the templates use the syntax \( ~ \) <TOP> or so) 02:24
lue
.oO( $data ~~ p/<ui32>**4/; $binfile.read(...) ~~ u/[ 0xFE 0xFF <BE> | 0xFF 0xFE <LE>] ... / )
Mouq s/"<TOP>"/"<~~>"/
Mouq wonders how hard it would actually be to get <~~> working 02:25
p6: say "a[bc[d]]" ~~ / \w+ [ \[ ~ \] <~~> ] / 02:26
camelia niecza v24-109-g48a8de3: OUTPUT«===SORRY!===␤␤Action method assertion:sym<~~> not yet implemented at /tmp/tmpfile line 1:␤------> say "a[bc[d]]" ~~ / \w+ [ \[ ~ \] <~~⏏> ] /␤␤Unhandled exception: Unable to resolve method oplift in typ…»
..rakudo-parrot 890878, rakudo-jvm 890878, rakudo-moar 890878: OUTPUT«Nil␤»
lue Despite my feelings that it's perhaps overkill, the most Perl 6-y option would appear to be a new sublanguage. 02:27
(method(s) on Buf is another nice way of doing it methinks)
Mouq p6: say "a[bc[d]]" ~~ / \w+ [ \[ ~ \] <~~> ]? /
camelia rakudo-parrot 890878, rakudo-jvm 890878, rakudo-moar 890878: OUTPUT«｢a[bc[d]]｣␤␤»
..niecza v24-109-g48a8de3: OUTPUT«===SORRY!===␤␤Action method assertion:sym<~~> not yet implemented at /tmp/tmpfile line 1:␤------> say "a[bc[d]]" ~~ / \w+ [ \[ ~ \] <~~⏏> ]? /␤␤Unhandled exception: Unable to resolve method oplift in ty…»
Mouq Rakudo++ :D
02:28 kbaker_ left
Mouq looks at learnyousomeerlang.com/starting-out-for-real "Bit Syntax!" 02:30
02:30 xragnar_ joined, xragnar is now known as Guest59407, Guest59407 left, xragnar_ is now known as xragnar
Mouq * learnyousomeerlang.com/starting-out...bit-syntax 02:30
lue The problem with this r-p failure is that the supposed diagnostics are all in formats like Int<7740517530316644326> :/ 02:32
jercos Mouq: would make base64 nicer anyway :p 02:33
Mouq The color thing at the link would be helped by Buf(:size) being implemented… 02:39
That way Erlang's "Color = 16#F09A29.␤Pixel = <<Color:24>>." could be written as "my $color = 0xF09A29;␤my $pixel = Buf.new(:size<24>,$color).size(8)" or something 02:42
bcode -82 02:44
er sorry :(
the first thing I say in here and it's a typoed command, what sort of impression that must leave...
lue .tell TimToady Thinking about pack/unpack, I can't tell if making it a sublang (with unpack easily seen as "binary regex", pack being something else entirely) is preferable, or if we'd rather toss (un)pack and have equivalent operations be done with Bufs (with an additional method or two on Buf as necessary). 02:46
yoleaux lue: I'll pass your message to TimToady.
lue bcode: don't worry about it :)
bcode yeah I tend to worry too much about things :) I've been mostly silently observing this channel so far as I think Perl 6 seems really cool, haven't had any time to do a lot with it so far due to IRL stuff though :( 02:47
Mouq Speaking of IRL stuff, I need to sleep
lue ♞ Mouq o/
bcode good night :) 02:48
Mouq bcode++ # observation :)
o/
dalek osystem: b3e975a | (Andrew Egeler)++ | META.list:
Add Net::POP3
 02:52
02:57 Mouq left 03:04 xenoterracide left
dalek rl6-most-wanted: 20f79c6 | (Andrew Egeler)++ | most-wanted/modules.md:
Add Net::SMTP and Net::POP3
 03:22
04:16 hoverboard joined 04:25 BenGoldberg left 04:30 SamuraiJack__ joined 04:33 thou left 04:34 Psyche^_ left 04:35 Psyche^ joined 04:52 rurban joined 04:59 bfulgham_ joined 05:02 hoverboard left 05:07 Visarga joined 05:09 Visarga left 05:26 bfulgham_ left
moritz \o #perl6 05:29
05:34 logie__ left
timotimo o/ 05:37
05:42 logie joined 05:46 aborazmeh joined 05:54 aborazmeh left 06:01 logie left
lue I'm randomly guessing this is some missing #?if parrot thing involving threads: gist.github.com/lue/9987997 06:02
♘ #perl6 o/
timotimo that's my guess, too 06:03
timotimo will have a look now
yes, the winners method ought to be if'd out 06:04
that seems to get the compilation going again 06:05
when it finishes i'll push
06:07 rurban left
timotimo dang, that wasn't it? 06:08
line: Int<-1172359952915819192> 06:09
yeah thanks
lue As you can see from my gist, the diagnostics are *super* helpful :) 06:12
lue really ♘
moritz seems like it's happening before the stringification of basic types is set up 06:13
and the one from Any is used
timotimo that's not so nice. should i try to write up a more daring failsafe for those types?
at least for Int? >_>
moritz for Int and Str at least 06:14
though the *really* interesting information here is in the unk_ hashes 06:15
timotimo yes, but the line number will get us so much closer
moritz (because that presumably tells us which symbols weren't found)
aye
timotimo now i'll just have to figure out where to find that code :P
possibly implement a method in Perl6::Compiler? 06:16
moritz it's in World
and that's where the stuff should go, IMHO
just git grep for 'Error while compiling' 06:17
method typed_exception
timotimo ah, yes, world
moritz maybe a method stringify_safely or so
timotimo i just had the presence of mind to grep for that 06:18
hm. how to best figure out if we're getting a "stupid" stringification? 06:19
except regex matching?
moritz just always do the safe thing for Str and Int
timotimo i suppose i could find_method and === against the find method for Any?
oh, that would work
moritz this code path is only triggered when compiling the setting 06:20
timotimo sounds sensible
i should be able to just unbox Str and Int and then nqp-stringify those, right?
06:21 kurahaupo joined
timotimo hmm. and i'll have to ~~ against find_symbol("Str") and "Int" respectively i believe 06:22
long time i've done anything in World.nqp
06:25 kaare_ joined
timotimo about to see if it works 06:28
line: 17784 06:30
better than nothing :)
but the line in this case is utterly useless, it just points at the very last lise
line*
hm. inside explain_mystery, "Parcel" should be available, shouldn't it? 06:41
06:41 molaf joined
timotimo it gets called at the end of parsing, so it *should* be there :\ 06:41
dalek kudo/safely_stringify_core_exceptions: 9cc033a | (Timo Paulssen)++ | src/Perl6/World.nqp:
attempt to stringify common structures for early comp exceptions

try to give better error messages than "Parcel<-1234134>" during compile-time error reporting when compiling the core setting		 06:44
timotimo if someone would like to figure out why this still gives Parcel<-1394879213847> instead of running my code, that'd be cool. i'm going to be AFK for a while 06:45
(fixing the underlying problem that causes the exception to happen in the first place would also be appreciated)
07:08 rurban joined 07:12 rurban left 07:24 araujo left 07:31 darutoko joined
moritz timotimo: if $Str && nqp::istype($target, self.find_symbol(["Str"])) { 07:35
timotimo: that won't work, because Str as a type object is always false 07:36
timotimo: also no need to find_symbol['Str']) again 07:37
timotimo: working on a fix now 07:39
07:39 bjz left, Rotwang joined
lizmat good *, #perl6! 07:39
checking out reported breakage now 07:40
moritz \o lizmat
07:40 bjz joined
lizmat moritz /o 07:40
moritz has to build a new NQP before being able to test his patch 07:41
07:42 pdcawley left
lizmat thinks she found the problem 07:44
.winners() should not exist in parrot, checking ifdeffed version now
dalek kudo/nom: 2be444c | (Elizabeth Mattijsen)++ | src/core/List.pm:
.winners should not exist in rakudo-p
 07:48
lizmat sorry for the inconvenience
Woodi hi today :) 07:49
moritz: you was mentioning some project with JSON messages, what is his name ?
also: what is 'actor semantic' ? :) 07:51
07:54 medatlantik joined 07:55 xfix left 07:56 xfix joined 08:02 medatlantik left
lizmat en.wikipedia.org/wiki/Actor_model 08:05
moritz Woodi: there's no name yet 08:07
I thought maybe "jirp" (tweet -> chirp, plus J from JSON), but that's already taken :( 08:08
dalek kudo/safely_stringify_core_exceptions: ef8d7bb | moritz++ | src/Perl6/World.nqp:
fix checks for wether type objects are available
08:09 rurban joined
moritz timotimo: perlpunks.de/paste/show/533fba3b.7ccc.21a that's the error message after my patch 08:10
timotimo: if you like it, feel free to squash-merge into nom
Woodi moritz: plan is to send views as html and data as json ?
08:11 denis_boyun_ joined 08:12 vendethiel left 08:13 rurban left 08:14 aborazmeh joined 08:16 woodvick joined
moritz Woodi: the plan is to send data as json, and format strings (for plain text) along 08:18
08:18 rindolf joined
moritz start simple, you know :-) 08:19
08:25 woodvick left 08:29 pdcawley joined 08:35 denis_boyun_ left 08:37 denis_boyun joined
Woodi actually I'm looking for something with fully static web pages presented to the user and only with artticles editors part with login form and js editor. something like jekyl but with edit part for not-devs, wysiwyg, then converted to static html... 08:39
08:57 kivutar joined 08:59 dmol joined 09:05 kurahaupo left, pdcawley left, kurahaupo joined, pdcawley joined 09:06 kaare_ left 09:09 rurban joined 09:12 fhelmberger joined 09:14 fhelmberger left, rurban left 09:17 Rotwang left 09:18 Rotwang joined
jnthn morning o/ 09:20
TimToady \o 09:23
yoleaux 02:46Z <lue> TimToady: Thinking about pack/unpack, I can't tell if making it a sublang (with unpack easily seen as "binary regex", pack being something else entirely) is preferable, or if we'd rather toss (un)pack and have equivalent operations be done with Bufs (with an additional method or two on Buf as necessary).
jnthn D'oh. Before wondering which changed made the NQP test suite take twice as long...check MVM_SPESH_LOG isn't defined in the env. 09:31
lizmat moin TimToady, jnthn
TimToady that makes it sleep like a log, eh?
lizmat
.oO( a very heavy speshial log )
 09:32
jnthn TimToady: Well, it makes it write a huge log file, so yeah, pretty much :)
TimToady well, they let us into the Forbidden City this morning, local
lizmat jnthn: did you ever envision using winner {} on multiple channels ? 09:33
TimToady not sure if our feet are still attached
jnthn TimToady: There's a good bit of walking to do there :) 09:34
jnthn was lucky enough to go there on a day when the sky was actually blue
TimToady esp if you do the park and lake to the northa as well
it was quite beautiful today
09:34 kurahaupo left
TimToady we've had really good air most days, or at least not terrible 09:35
jnthn lizmat: Yes, though when I envisioned it it was called select, like in Go. But yes, the point of it is you can be handling a bunch of channels.
lizmat so how do you handle done @channels ?
you cannot simply last on the the first "done"
jnthn TimToady: ah, that's good. My last trip, the air was pretty awful. The trip before it was better and I saw plenty more ;)
lizmat but a next iteration may select a "done" of a previously closed channel, so using a counter doesn't work 09:36
jnthn lizmat: I guess you need to maintain a "to check" list. 09:37
lizmat: And remove those you invoked done on.
sergot hi! o/
lizmat jnthn: outside of the winner {} or inside ?
winner {} is basically stateless
jnthn Ah... 09:38
Yeah, now I see the problem.
lizmat so it would become the burden of the developer
jnthn ugh
lizmat yes, indeed
jnthn That isn't the design I was aiming for at all...
Hmm.
lizmat indeed... it feels to me that winner {} needs some state somehow
that would also be good from a performance point of view :)
jnthn Well, probably need to put together some use cases... 09:39
lizmat more @c { @a.push: $:v } 09:40
done @c { @a.push: "done $:k"; last if ++$done == 2 }
is what I tried to put in a test
of course, it doesn't work, as it may end prematurely
also, in the current implementation, the $:k on the done {} doesn't make sense, but that's an implementation detail
09:40 aborazmeh left
jnthn I'm not too enthralled with the named params here either fwiw. 09:41
Because winner then iiuc needs to do introspection.
lizmat the more I think about this, using winner {} for both promises and channels and potential mixes
is confusing
at least in the current spec 09:42
jnthn I'm not so sure. I know we can define a Promise.Channel soundly, so I think the equivalence is good enough.
The problem we're seeing here with done is creatable entirely with channels.
lizmat indeed... but we're using "done" in the Promise context to return a value, and "done" in the Channel context to signify the channel is closed 09:43
maybe we need to call "more" on the Promise as well ?
09:44 aborazmeh joined
lizmat trying to write extensive tests for winner {} and .winners, puts my nose deep into these things 09:45
and it smells :-)
jnthn Oh...I'd expect "more" to be always about values, "done" to always be about completion with no interest in a value.
Otherwise we can't say that winner on a promise is the same semantically as if you were doing winner on $that_promise.Channel.
Which I think we should be able to.
But yeah, something is off here. 09:46
lizmat there's actually an error now to do "more" on a Promise 09:47
jnthn ugh
lizmat elsif $kind == $WINNER_KIND_MORE && $next ~~ Promise {
die "Cannot use 'more' on a Promise";
jnthn But "more" doesn't really convey the right thing on a Promise.
lizmat ok, let me simmer about this offline some 09:48
well, yeah...
jnthn I think the whole construct needs a revisit.
lizmat indeed, I'm coming to that conclusion as well
in a way, I was doing this with .winners, but I guess we need to go deeper 09:49
09:51 aborazmeh left 09:59 aborazmeh joined
dalek ast: 65ddc70 | (Elizabeth Mattijsen)++ | S17-concurrency/winner.t:
Some more winner {} tests

Which just show we need something better specced and implemented		 10:03
10:06 spider-mario joined
TimToady how does .winner differ from "race" 10:07
10:07 aborazmeh left
TimToady er, .winners 10:07
10:08 Rotwang left
TimToady doesn't like calling all of the finishers "winners", political correctness notwithstanding 10:08
lizmat because .winners returns value from multiple channels / promises
and race just changes the order in which an iteration is done 10:09
TimToady still don't much like the name
10:09 denis_boyun left
lizmat I'm not happy about the name either and open to suggestions :-) 10:09
10:10 aborazmeh joined
lizmat for a while, I thought about calling it .process 10:10
TimToady doesn't have a good name at the moment, but will think about it
lizmat but that was too generic
10:10 rurban joined 10:11 denis_boyun joined
TimToady it's a kind of merge in the time dimension 10:11
10:12 SamuraiJack__ left
lizmat with nothing happening simultaneously anymore 10:12
it's also a serializing of events in time
TimToady well, that's what races do, determine order of finish
lizmat oddly enough, "race" is currently specced as "I don't care in which order they finish" 10:13
TimToady welll, in a sense, the race organizers don't care in advance either :)
lizmat true 10:14
TimToady @dostuff.athon :)
and marathons are famous for letting everyone finish 10:15
lizmat athon { } instead of winner { }
10:15 rurban left
TimToady no, winner is specifically one one 10:15
that's why it has that name
lizmat athon {} instead of .winners()
TimToady to preclude people thinking it's all of them
that makes more sense to me 10:16
lizmat so athon { } would return a lazy list of the values it received
10:16 kivutar left 10:17 aborazmeh left
TimToady is not sure it makes sense semantically to mix promise values in with channel values like that, but is open to persuasion 10:17
10:17 SamuraiJack__ joined
lizmat still, if we have athon {}, I'm not sure the use for winner P} would be 10:17
TimToady it feels like a type failure
lizmat well, that too
TimToady winner would shoot all the other contestants 10:18
lizmat but you could consider a Promise to be a Channel only returning a single value
TimToady you could, but then you get mixed up with the promise that a chennel provides, I suspect
lizmat hmmm.. the Promise a Channel provides? 10:19
10:19 aborazmeh joined
spider-mario hm, on Rakudo 2014.03 on MoarVM, just typing “@x” generates an infinite loop of error messages 10:20
TimToady I suppose we could say that done might or might not provide a final value; I could see some channel algos that would want to work that way
moritz spider-mario: iirc that was fixed after the release
spider-mario oh, nice
thanks
TimToady but there was some reaons I didn't level promises and channels that I'm too tired to recall at the moment 10:21
might've been a dumb reason, but I did consider it, that much I remember :)
lizmat :-)
TimToady well, we need to go find dinner before all the restaurants get even more overfull than they are now... 10:22
lizmat bon appetit! 10:23
jnthn argh 10:26
uh, ww
10:26 aborazmeh left 10:35 aborazmeh joined 10:38 SamuraiJack__ left
Woodi winners are like #pragma omp parallel for with collecting array being fifo queue ? 10:40
lizmat sounds like it, not sure :-) 10:50
Woodi such patters: map-paraller_burst-collect probably are common case but rather actorless... and example is to simple to define graph of parallel nodes with topologies... but methink we still not quite think in sequential parallel proces_sess_. something like: BEGIN { start helper, collector, io; } 11:03
11:12 aborazmeh left
spider-mario rakudo star, even with Configure.pl --backends=parrot, appears to build the MoarVM backend when nqp-m is present 11:19
well, perhaps not all of it 11:20
at least /usr/bin/perl6-m and /usr/languages/perl6/runtime/perl6.moarvm for example
jnthn Hm, interesting. :) 11:22
I guesss maybe it doesn't pass a --backend down to Rakudo's Configure
spider-mario: Are you talking about the last star release, or if you build one from the current star repo?
spider-mario 2014.03
jnthn Ah. I guess that's it, then. 11:23
The 2014.04 should support backends properly, so should be fixed there...in theory. 11:24
spider-mario ok, thanks :) 11:25
I guess I can just grab a patch from the appropriate commit in the meantime, then
11:28 dwarring left
spider-mario I don’t think it’s fixed 11:38
the culprit appears to be line 53 of tools/build/Makefile.in 11:39
cd $(RAKUDO_DIR) && $(PERL) Configure.pl --prefix=$(PREFIX_DIR) && $(MAKE)
(line 53 as of star 2014.03-33-gf79eaed) 11:40
11:40 grondilu joined
spider-mario (commit f79ea) 11:40
grondilu my naive attempt to make a parallel mandelbrot: gist.github.com/grondilu/9990847 11:41
spider-mario the last time this line was touched was back in november
jnthn Ah, yeah. Looks like it needs patching to pass the backends down. 11:44
11:44 mkz left
colomon grondilu: does that work on moar? 11:50
grondilu colomon: not really. It does not behave deterministically, for one. 11:51
11:51 mkz joined 11:53 ilbot3 left 11:54 ilbot3 joined
colomon grondilu: does it work on jvm? 12:00
(ie is it naive because of a lack of sophistication or because it doesn't work?)
(or in other words, what do I need to do to run the script? ;)
dalek kudo/nom: aab600f | (Timo Paulssen)++ | src/Perl6/World.nqp:
Stringify common structures for early comp exceptions

try to give better error messages than "Parcel<-1234134>" during compile-time error reporting when compiling the core setting
Contains some small fixups by moritz		 12:02
moritz timotimo: ok, I was impatient, and squash-merged the branch myself
spider-mario is MoarVM bytecode architecture-dependant? 12:06
grondilu colomon: it's naive because it's my first attempt and because it's not sophisticated. I have no idea if it's supposed to work right now on Moar. But it does seem to run without too much complain.
sometimes it interrupts with weird errors like: 12:08
Unhandled exception: When invoking EXCEPTION, Provided outer frame 0x9632ea0 (MVMStaticFrame postcircumfix:<[ ]>) does not match expected static frame type 0x8b6d510 (MVMStaticFrame )
I say it's weird because it seems to do that in the middle of the drawing. Though I should check that. 12:10
12:12 rurban joined
jnthn spider-mario: No 12:13
spider-mario ok, thanks.
jnthn spider-mario: Though the VM itself certainly has its share of porting work to go. 12:14
grondilu colomon: haven't tried on jvm 12:15
grondilu does not use the jvm
12:17 rurban left 12:22 d0nk left, d0nk joined 12:34 ilbot3 left, ilbot3 joined 12:38 mkz left 12:40 kivutar joined 12:50 baest left 12:56 baest joined 13:00 baest left 13:01 baest joined 13:03 xfix left 13:05 jdv79 joined, mkz joined
jdv79 where is "#?endif" parsed? 13:10
13:11 xfix joined 13:12 baest left
jnthn tools/build/gen-cat.pl iirc 13:12
13:13 rurban joined 13:14 baest joined, dayangkun_ left 13:17 rurban left 13:20 aborazmeh joined, xfix left 13:23 xenoterracide joined, xfix joined 13:28 xfix left 13:31 xfix joined 13:38 mtk left 13:39 hummeleB1 joined 13:41 mtk joined 13:42 colomon left 13:44 grondilu left 13:45 baest left 13:50 baest joined 13:56 baest left 13:59 rurban joined 14:00 baest joined 14:01 mtk left 14:06 mtk joined, telex left 14:08 telex joined
FROGGS morning 14:21
14:23 logie joined 14:25 vendethiel joined
masak morning, FROGGS 14:30
14:33 Rotwang joined
dalek p/execname: 44181ad | jonathan++ | src/NQP/Optimizer.nqp:
Start recording var decls/usages.
 14:37
p/execname: d062aa3 | jonathan++ | src/NQP/Optimizer.nqp:
Implement lexical => local lowering.
p/execname: a83cec1 | jonathan++ | src/NQP/Optimizer.nqp:
Can lower $/ and self also.
p/execname: 39e7a00 | jonathan++ | src/NQP/Optimizer.nqp:
Fix spello; [Coke]++.
p/execname: 4dd7814 | jnthn++ | src/NQP/Optimizer.nqp:
Walk default defaults in the optimizer.
p/execname: d317008 | jnthn++ | src/vm/jvm/ModuleLoader.nqp:
Eliminate epic hack in JVM ModuleLoader.

The optimizer now got clever enough to see through the hack.
p/execname: dcb3c38 | (Tobias Leich)++ | src/ (2 files):
Merge branch 'master' of github.com:perl6/nqp into execname
jnthn ooh, free karma! 14:39
FROGGS well, almost free :o) 14:47
you need to do something in the first place 14:48
14:48 gtodd joined
moritz FROGGS: did you see my note yesterday abot execname? 14:54
FROGGS no, let me backlog
moritz irclog.perlgeek.de/perl6/2014-04-04#i_8543190
FROGGS moritz: that sounds good 14:57
but, do we really want to keep the shell wrappers?
moritz no idea 14:58
gtodd Is it true a long time ago people were able to get parrot run both perl5 & perl6 15:00
moritz gtodd: that was always the plan, but it never worked.
gtodd if so this would me we can haz ponies one day? i.e. in the spirit of www.poniecode.org
15:00 kaare_ joined 15:01 zakharyas joined
gtodd moritz: interesting that it got as far as it did ... parrot seems more capable now than in 2006 15:01
so perhaps one day we can haz 15:02
moritz not at parrot's current rate of development
15:02 xenoterracide left 15:03 xenoterracide joined
jnthn More likely to see that happen through one of v5 or some MoarVM-based itnerop project, I'd say. 15:03
15:04 pecastro left
moritz www.ohloh.net/p/parrot take a look at the "Activity" section 15:05
gtodd $GIANT_CORPORATE_SPONSOR ??
moritz manages not to joke about dead horses and dead parrots 15:06
FROGGS *g*
gtodd hmm well ohloh shows a levelling off ... :-) 15:07
FROGGS moritz: about the runner: I am going to play with embedded mbc bytecode now, and if that works out there would be a nice reason to use this kind of runner
moritz: if not, we can always keep the shell wrapper with a --execname...
jnthn FROGGS: Note that the shell script runner did have the advantage of easy relocatability. :) 15:08
FROGGS: 'cus relocating would be "change paths in the shell script"
FROGGS jnthn: true
but the to create a new nqp-m in another place you'd just create the gen-runner thing again 15:09
timotimo the runner could provide a bit of extra space for the paths and an argument that causes it to read in its own code, re-write the paths and spit out a "relocated" runner :P
FROGGS O.o 15:10
timotimo you know how zip files can be appended to any file you like and still work properly? 15:11
jnthn 'cus it looks at the end of the file first for the "header", iirc? :)
gtodd I guess someone needs to convince the people who have budgets for it why sponsoring parrot (and its relatives i.e. rakudo) is cool and will position them to *take over the cloud* 15:12
timotimo yes and the references from the header are probably relative rather than absolute
gtodd oops I think I meant "world" not "cloud" .. but cloud sounds cooler so ...
15:13 colomon joined
gtodd anyway while you guys keep working perl6 keeps getting more and more fun easier and faster to play with so thank you 15:13
15:14 MrMeek left, treehug88 joined, treehug88 left 15:15 treehug88 joined
timotimo gtodd: well, you can probably see we're all mostly excited about MoarVM these days, rather than parrot 15:16
moritz FROGGS: instead of reserving space for relocation paths, the fake-executable could know how to extract its payload, and create a new, relocated copy of itself 15:20
FROGGS moritz: what is a "fake-executable" here?
the shell wrapper? 15:21
moritz FROGGS: no, the moar exec + byte code embedded
FROGGS ahh, but that is not a fake executable for me... but perhaps I just don't know the diff between a fake and a normal executable :o) 15:22
timotimo hah. 15:23
15:25 aborazmeh left
timotimo implementing andthen as well as orelse properly on rakudo doesn't seem terribly hard 15:30
maybe i'll do that next, just to remove a yellow from the compilers/features comparison matrix :P
15:32 colomon left
timotimo huh. there already is a "andthen_op" piece of code that at least block-closure-thunk-refs the RHS of each of the andthens 15:34
that's a good starting point, i'd bet
15:34 IllvilJa left 15:35 raiph joined
timotimo hm. the spec doesn't say anything about "exception semantics" for andthen 15:36
15:36 guru joined, guru is now known as Guest52690
timotimo ah, exception semantics only "exist" for orelse 15:36
timotimo heads out again for a while
15:36 Guest52690 is now known as ajr_
moritz timotimo: it sets $! in some weird scope, or something 15:42
vendethiel r: gather { take $a; take |@b; }; 15:49
camelia rakudo-jvm aab600: OUTPUT«(timeout)»
..rakudo-parrot aab600, rakudo-moar aab600: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Variable '$a' is not declared␤at /tmp/tmpfile:1␤------> gather { take $a⏏; take |@b; };␤ expecting any of:␤ postfix␤»
vendethiel r: my $a = 1; my @b = (2, 3); say (gather { take $a; take |@b; }).perl; 15:50
camelia rakudo-parrot aab600, rakudo-jvm aab600, rakudo-moar aab600: OUTPUT«(1, (2, 3)).list␤»
vendethiel r: my $a = 1; my @b = (2, 3); say (gather { take $a; take @b; }).perl;
camelia rakudo-parrot aab600, rakudo-jvm aab600, rakudo-moar aab600: OUTPUT«(1, 2, 3).list␤»
vendethiel r: my $a = 1; my @b = (2, 3); say (gather { take $a; take $@b; })[1].perl; 15:51
camelia rakudo-parrot aab600, rakudo-jvm aab600, rakudo-moar aab600: OUTPUT«[2, 3]␤»
15:56 IllvilJa joined, hoverboard joined 15:57 treehug88 left, mtk left
timotimo moritz: only orelse does, andthen doesn't apparently 16:07
in any case, there ought to be some tests and it seems like there are none
or at least none that are linked to the synopses
16:23 pecastro joined
lizmat is back from cycling 16:24
jnthn: woolfy came up with a better name for winner {} and .winner 16:25
outcome
outcome @p,@c
would give you a lazy list with results from promises and channels until all are kept / closed 16:26
jnthn Hmm :)
lizmat outcome @p,@c { ... }
would execute the code whenever a promise was kept / channel produced a value
outcome * { done / more } 16:27
would have the same syntax as now winner {}
in this scenario, I don't really see a reason why "winner {}" or .winners() should exist
16:29 thou joined
lizmat my @a := outcome ... # turn into an array so you can carry it "around" 16:32
in that context, I wonder if we could use a special type of array 16:33
which, when doing something like:
for @a { last if $foo }
would continue where it left off if you would do a
for @a {...}
again
if so, you could do something like: my @a := outcome { wait 0 { last } }; loop { do something; for @a { process $_ } } 16:36
jnthn: does that make sense ? 16:37
jnthn lizmat: Does the array case imply some kind of implicit loop, which winner lacks today? 16:38
lizmat outcome returns a lazy list, so yes 16:39
winner {} either needs state, or it needs an internal loop
the way it is specced now, makes it virtually useless for any multi-promise and/or multi-channel cases 16:40
s/it/winner {}/
as there is a good chance you will see the same promis again and again, unless you remove it from winner {} yourself 16:41
which is a lot of bookkeeping :-(
jnthn yeah, agree. 16:42
lizmat I'll probably test an implementation of outcome {} later tonight, and if it works, spec it, ok? 16:45
16:46 denis_boyun_ joined, denis_boyun left
dalek ast: 3b5223c | (Elizabeth Mattijsen)++ | S17-concurrency/lock.t:
Make lock test failure a bit more verbose
 16:50
lizmat jnthn: see above change 16:51
about 1 in 15 times I get:
not ok 7 - Condition variable worked
# got: 'ale,porter'
# expected: 'ale,porter,stout'
# log = ale porter stout
raiph lizmat: note how you used the word "whenever" which I know at least one other person liked 16:52
16:52 kurahaupo joined
raiph whenever * { done / more } 16:53
16:53 denis_boyun_ left
lizmat but that would just be a synonym for "winner", no? 16:53
raiph winner implies one thing 16:54
lizmat and my point is that the current semantics of "winner {}" are not of much use in real life
raiph whenever (to me) implies, well, whenever something happens
16:54 denis_boyun_ joined
lizmat which is what "winner {}" now does 16:54
raiph including when there are a series of things that happen 16:55
lizmat my proposal for "outcome" would be something with the syntax of "winner {}" (more or less) with the semantice of .winners()
raiph you said it yourself: "would execute the code whenever a promise was kept / channel produced a value "
vendethiel r: say @$@$(1, 2).perl;
camelia rakudo-parrot aab600, rakudo-jvm aab600, rakudo-moar aab600: OUTPUT«(1, 2).list␤» 16:56
timotimo vendethiel: time for swearword-multigils
raiph right, i'm bikeshed suggesting "whenever" as an alternate to outcome, not to winner
lizmat ah, ok
whatever :-)
seriously
raiph heh
vendethiel r: say @%$%@$(1, 2).perl # trying to spice it up for timotimo
camelia rakudo-parrot aab600, rakudo-jvm aab600, rakudo-moar aab600: OUTPUT«("1" => 2,).list␤»
lizmat whenever feels more like a single event handler to me
vendethiel that'll work.
timotimo rrrrrr spicy 16:57
lizmat outcome feels more like with a bicycle race, the list of who got in first, second, third all the way to the last 16:58
raiph lizmat: then why did you say "whenever" in explaining "outcome"?
lizmat it was about the code block specified with outcome @p,@c { ... }
another word for "outcome" that I came up with, was "mapathon"
because it is basically a map {} 16:59
with the promises being kept/broken and values from channels becoming part of a lazy list
until there were no unPlanned Promises left, and all channels were closed 17:00
raiph k
lizmat so, just calling "outcome {}" would block in the current thread 17:01
retupmoca Could I get someone to glance at github.com/rakudo/rakudo/pull/255 ?
I also have a couple of open pull requests for panda, if someone here has time and a commitbit
raiph lizmat++ # awesome work on P6 for last year+ 17:02
xenoterracide I swear the feature I want in every language more than anything in the world, is making it so that that by default, methods cannot take null/undef parameters. 17:03
lizmat raiph: *blush* :-)
in Perl 6: make sure your methods have the right signature :-) 17:04
retupmoca xenoterracide: it's not by default, but you can do this:
m: sub foo($x where .defined) { say $x }; foo("ASDF"); foo(Str)
camelia rakudo-moar aab600: OUTPUT«ASDF␤Constraint type check failed for parameter '$x'␤ in sub foo at /tmp/Kjn2bXKsSD:1␤ in block at /tmp/Kjn2bXKsSD:1␤␤»
xenoterracide retupmoca: you can do :D 17:05
lizmat retupmoca: since FROGGS is working in this area, I'm a bit uncertain to merge that PR
xenoterracide retupmoca: but the most frequent error I see is forgetting to define something and forgetting a null/undef check
lizmat afk for a few hours& 17:06
retupmoca lizmat: no worries, I'll wait for him to show up / notice the PR
xenoterracide maybe TimToady would consider making :D the default for method parameters, instead of :U (notes that read only would be nice too but less required). (Only bringing this up because I don't believe it would be impossible for perl6 to change this yet ) 17:10
vendethiel r: my $a = 'foo'; say ($a // '?' ~ 'b') 17:13
camelia rakudo-parrot aab600, rakudo-jvm aab600, rakudo-moar aab600: OUTPUT«foo␤»
17:17 denis_boyun___ joined 17:19 denis_boyun_ left 17:24 molaf left 17:39 bjz left, bjz joined 17:40 pecastro left 17:54 leosik joined
dalek kudo/nom: 2de2c9f | (Andrew Egeler)++ | src/core/terms.pm:
Load modules from %CUSTOM_LIB<home> before <site>

This way a user can install a local module that will override a system-installed one		 17:56
kudo/nom: a667490 | (Tobias Leich)++ | src/core/terms.pm:
Merge pull request #255 from retupmoca/nom

Load modules from %CUSTOM_LIB<home> before <site>
retupmoca FROGGS++ # making my itch-scratching official 18:00
FROGGS pleasure :o) 18:01
18:05 xinming__ left 18:06 treehug88 joined 18:07 xinming_ joined 18:10 treehug88 left 18:11 hoverboard left 18:25 pecastro joined
timotimo r: try die "hello there"; say ~$! eq "hello there" 18:26
camelia rakudo-parrot aab600, rakudo-jvm aab600, rakudo-moar aab600: OUTPUT«True␤»
timotimo r: sub output_error { say $! }; try die "hey"; output_error; 18:32
camelia rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«Nil␤» 18:33
timotimo isn't $! supposed to be dynamic?
moritz n: sub output_error { say $! }; try die "hey"; output_error 18:36
camelia niecza v24-109-g48a8de3: OUTPUT«(Any)␤»
vendethiel timotimo: I don't think so (outside the fact it'd be $*!, it'd make CATCH and all be dynamic too) 18:37
timotimo hm 18:39
so i'll probably have to fiddle around with the blocks that get created if they're not pointy 18:41
18:43 leosik left 18:46 rurban left 18:49 darutoko left 18:54 kaare_ left 18:58 xinming_ left 19:00 xinming_ joined
timotimo r: say 1 // "foo" // say "hi" 19:02
camelia rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«1␤» 19:03
timotimo oh, that does thunk
19:05 LLamaRider joined
dalek ast: 580f5c1 | (Timo Paulssen)++ | S03-operators/orelse.t:
a few tests for the orelse operator

a bit of sanity and then exception semantics.		 19:06
kudo/nom: c2cc4fd | (Timo Paulssen)++ | t/spectest.data:
run the new orelse tests
19:06 molaf joined
timotimo i see an opportunity for a tiny optimization in rakudo's EXPR; it has a bunch of if/elses in a row that all begin with "!$past &&" 19:12
i wonder if factoring that $!past out would be worth a lot
oh and btw, is there some good document on how to understand how adverbs attach to operators? 19:13
i seem to recall a not quite simple explanation involving precedence and stuff
if there isn't, i may write one up and put it $somewhere
vendethiel timotimo: please do :) 19:14
tadzik omg, internet
timotimo tadzik: \o/ 19:16
how are you today?
tadzik a bit ache-y, but fine aside from that :) 19:20
I'm like in the middle of a 26 bus ride back home 19:21
timotimo 26 different bus lines?
tadzik oh, sorry: 26-hour long
moritz woah
tadzik I wish I was kidding :P 19:22
but at least I'm now in a country that actually has internet
vendethiel r: Any or Int or 1
camelia ( no output )
vendethiel r: say Any or Int or 1 # :(
camelia rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«(Any)␤»
19:22 SamuraiJack__ joined
vendethiel r: say Any orelse Int orelse 1; say Any // Int // 1; 19:23
camelia rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«(Any)␤1␤»
moritz r: say (Any or Int or 1)
vendethiel r: say (Any orelse Int orelse 1); say Any // Int // 1;
camelia rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«1␤»
rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«1␤1␤»
moritz vendethiel: precedence matters :-)
vendethiel moritz: got it ;)
I just don't understand "orelse" semantics 19:24
r: say try die "oh noes" or 1;
camelia rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«Nil␤»
vendethiel r: say try die "oh noes" orelse 1;
camelia rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«Nil␤»
vendethiel r: say try (die "oh noes" orelse 1); 19:25
camelia rakudo-parrot a66749, rakudo-jvm a66749, rakudo-moar a66749: OUTPUT«Nil␤»
vendethiel uh yeah it's as a todo right now ;_;.
So basically, `orelse` treats exception like `or` treats falsy values ? I remember looking at synopsis but I didn't get it 19:26
tadzik in the backlog someone said "tadzik is at fault :o" 19:29
timotimo vendethiel: orelse handles "undefined" values
vendethiel timotimo: how's it different to // ?
timotimo and passes them to all parameters of the next thing if it's a pointy block, otherwise creates a $! to bind to
hm.
r: say die "what is going on!" // 1 19:30
camelia rakudo-jvm c2cc4f: OUTPUT«(timeout)»
..rakudo-parrot c2cc4f, rakudo-moar c2cc4f: OUTPUT«what is going on!␤ in block at /tmp/tmpfile:1␤␤»
timotimo orelse wouldn't sink this exception
vendethiel timotimo: your tests include a `try`; though
timotimo it's so that the whole test file doesn't blow up if the exception gets sunk :) 19:31
vendethiel Alrighty. I'm not sure why it's doing `or`'s job too, tho ?
timotimo you mean it should only go to the next thing if there was a failure? 19:32
rather than any undefined value?
synopsis says "[orelse] Returns the first argument that evaluates successfully (that is, if the result is defined)." 19:33
vendethiel alrighty. That's a bit hard to grasp if you don't connect "successfully" with "without an exception" 19:34
timotimo right.
vendethiel so, you could add CATCH { fail; } after the `try`s in your last commit test file ?
timotimo fail?
vendethiel fail a test 19:35
`try { die "oh noes" orelse $result = 1; CATCH { $result = 0; } }; ok $result;`
timotimo r: use Test; fail "whhyyyyyyyy"
camelia rakudo-parrot c2cc4f: OUTPUT«whhyyyyyyyy␤current instr.: 'throw' pc 418065 (src/gen/p-CORE.setting.pir:174958) (gen/parrot/CORE.setting:11009)␤called from Sub 'sink' pc 456246 (src/gen/p-CORE.setting.pir:189173) (gen/parrot/CORE.setting:12446)␤called from Sub 'MAIN' pc 379 (sr…»
..rakudo-jvm c2cc4f: OUTPUT«Unhandled exception: whhyyyyyyyy␤ in throw (gen/jvm/CORE.setting:11008)␤ in sink (gen/jvm/CORE.setting:12444)␤ in MAIN (gen/jvm/main.nqp:47)␤ in (gen/jvm/main.nqp:41)␤ in (gen/jvm/main.nqp)␤␤»
..rakudo-moar c2cc4f: OUTPUT«Unhandled exception: whhyyyyyyyy␤ at src/gen/m-CORE.setting:11067 (/home/p6eval/rakudo-inst-1/languages/perl6/runtime/CORE.setting.moarvm:throw:129)␤ from src/gen/m-CORE.setting:12501 (/home/p6eval/rakudo-inst-1/languages/perl6/runtime/CORE.setting…»
timotimo fail is actually not a Test thing
i think you mean flunk or something?
r: use Test; flunk "whyyyy" 19:36
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«not ok 1 - flunk whyyyy␤»
vendethiel timotimo: yeah :). "carp" :p
(at least I think carp is a perl thingie ?)
timotimo that's what i use the result thingie for at the moment
carp is a perl5 thing at least
vendethiel timotimo: would the code I posted just before your "r: fail why" work or not ? 19:37
(in a 100% rakudo world)
timotimo i think it would 19:38
but i don't see why i would have to put the catch in
vendethiel to check the exception is correctly "stopped"
I'm not saying you should put it, I'm just asking for correct behavior 19:39
(check that orelse doesn't rethrow like some mechanics in some languages...)
lue Couldn't you use something like dies_ok ? 19:40
vendethiel wouldn't dies_ok check if it indeed threw an exception ? 19:41
timotimo lives_ok, actually
vendethiel then yeah
timotimo: wouldn't it make sense to have `lives_ok { die "oh noes" orelse $result = 1 }` ? 19:43
timotimo sounds good to me actually
if you'd like to patch it, feel free :)
vendethiel timotimo: should I patch every try, or only once ? 19:44
timotimo every one of them
should result in cleaner code
vendethiel timotimo: hum, how so? 19:45
timotimo ... it would not?
vendethiel I'm just thinking of adding `lives_ok ` before each `try`. Not sure why that's cleaner?
well before code blocks
r: use Test; lives_ok { die "oh noes"; } 19:46
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«not ok 1 - ␤# oh noes␤»
vendethiel r: use Test; lives_ok { try die "oh noes"; }
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«ok 1 - ␤»
vendethiel timotimo: definitely cleaner code
19:47 rurban joined
vendethiel timotimo: github.com/perl6/roast/pull/39, is that okay ? 19:47
timotimo could you give the lives_ok a little message, too? 19:48
vendethiel r: lives_ok { die "oh noes" orelse ok True, "orelse works!"; }, "orelse sinks exceptions"; 19:50
camelia rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Undeclared routines:␤ lives_ok used at line 1␤ ok used at line 1. Did you mean 'on'?␤␤»
..rakudo-parrot c2cc4f: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Undeclared routines:␤ lives_ok used at line 1␤ ok used at line 1␤␤»
vendethiel r: use Test; lives_ok { die "oh noes" orelse ok True, "orelse works!"; }, "orelse sinks exceptions";
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«not ok 1 - orelse sinks exceptions␤# oh noes␤»
19:51 hoverboard joined, rurban left
timotimo oh, something i forgot about 19:51
this will give variable numbers of tests depending on how it acts 19:52
vendethiel lives_ok should count for 2 tests, basically ?
there's no "plan", though 19:53
19:53 pdcawley left 19:54 pdcawley joined
vendethiel (an exception that sinks is an exception that gets thrown ? "method sink() { $!exception.throw unless $!handled }". Thought it was the opposite, as I'm not sure of the overloaded meanings of "sink") 19:54
timotimo well, there may not be a plan, but diffing runs where something in the middle changes will be *weird* 19:55
sink basically means something like "doesn't get assigned to a variable or returned" 19:56
19:57 raiph left
vendethiel so the message should be "orelse doesn't sink exception" ? 19:57
(I get "sink context" for values; not so much for exceptions) 19:58
Well, not sure how to get the correct test count then -- I think there's a way to make a test count as two passed? 20:00
20:00 rurban joined
timotimo hmm. maybe something like "orelse doesn't rethrow exception" would be better? 20:00
well, the exception causes the "ok" to be skipped
vendethiel yeah, that I get. 20:01
timotimo the try i had in it before prevented that exact problem :\
vendethiel I can use the CATCH method I talked about earlier 20:02
20:02 zakharyas left
timotimo hmm, CATCH { flunk "exception rethrown" } ? 20:02
vendethiel flunk just fails instead of being counted as a failed test ? 20:03
CATCH { $result = 0 } ? I thought about flunk; but that's one less test if it succeeds as well, no ?
timotimo i think flunk is like "ok 0" 20:04
20:04 rurban left
vendethiel but that's an `ok` a correct `orelse` will never hit, right ? 20:05
timotimo yes 20:07
so the correct orelse will reach the "ok $result"
and an incorrect one will get the flunk instead
vendethiel an incorrect orelse will get to the `ok` too, because of the try. 20:08
unless you want me to move the try "outside"?
r: use Test; { my $result; try { die "oh noes" orelse $result = 1; CATCH { flunk "test failed!" } }; ok $result, "orelse works"; } # incorrect orelse will get to the `ok` 20:09
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«not ok 1 - flunk test failed!␤oh noes␤ in block at /tmp/tmpfile:1␤ in block at /tmp/tmpfile:1␤␤»
timotimo the try can be replaced with lives_ok in that case 20:10
20:11 SamuraiJack__ left
vendethiel timotimo: `try { die "oh noes" orelse ok 1; CATCH { flunk "doesn't work" } }` # does it thunk ? 20:12
timotimo orelse is supposed to thunk and i believe it already does in rakudo 20:13
vendethiel so would that work ?
timotimo because currently it uses the "defor" op internally so far
well, if it is wrong in the way that it won't thunk, we'll get an ok in this case :)
though it'll show up as a not ok in an earlier test 20:14
vendethiel timotimo: is that implem. okay for you ? 20:19
timotimo i suppose it can be okay 20:20
so in order to implement the correct thunking for orelse, i copied "make_topic_block_ref" to "make_exception_topic_block_ref", but i need to mix that with make_thunk_ref and then block_closure 20:25
sadly, make_thunk_ref uses create_simple_code_object, which creates a block without a signature, which is wrong for my purposes 20:26
a tip on how to avoid code duplication here?
can i twiddle the signature later on perhaps?
20:27 LLamaRider left
vendethiel r: use Test; diag "ok" 20:29
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«# ok␤»
vendethiel diag's basically flunk's opposite?
r: try { die "a"; CATCH { say "foo"; } } 20:31
timotimo well ... for some value of opposite
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«foo␤a␤ in block at /tmp/tmpfile:1␤␤»
timotimo it outputs the text you give it as a TAP-aware comment
vendethiel r: try { die "a"; CATCH { default { say "foo" } } }
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«foo␤»
vendethiel r: try { die "a" orelse diag("orelse works !"); CATCH { flunk "orelse shouldn't rethrow exceptions" } } 20:32
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Undeclared routines:␤ diag used at line 1␤ flunk used at line 1. Did you mean 'link'?␤␤»
vendethiel r: use Test; try { die "a" orelse diag("orelse works !"); CATCH { flunk "orelse shouldn't rethrow exceptions" } }
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«not ok 1 - flunk orelse shouldn't rethrow exceptions␤a␤ in block at /tmp/tmpfile:1␤␤»
vendethiel timotimo: ^^ basically what I'm getting at
timotimo diag doesn't count as a passed test
vendethiel oh yeah, it doesn't call proclaim 20:33
I guess I'm looking for "pass"
r: use Test; try { die "a" orelse pass "orelse works !"; CATCH { flunk "orelse shouldn't rethrow exceptions" } } 20:34
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«not ok 1 - flunk orelse shouldn't rethrow exceptions␤a␤ in block at /tmp/tmpfile:1␤␤»
timotimo oh, yes, pass
vendethiel timotimo: I currently have `{ lives_ok { ... } }`. Should I remove outer brackets or does that make #? comments fail ? 20:35
20:35 klapperl_ joined
timotimo keep them, for that exact reason 20:35
20:36 klapperl left
vendethiel guessed so 20:36
r: use Test; class LilStr { method Str { "foo" } }; my $lil = LilStr.new; ok ~$lil eq "foo"; 20:37
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«ok 1 - ␤»
20:37 hoverboard left
vendethiel r: use Test; class LilStr { method Str { "foo" } }; my $lil = LilStr.new; ok ~$lil eq "foo", "It does work !" 20:37
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«ok 1 - It does work !␤»
vendethiel timotimo: updated 20:41
timotimo i *think* it's all right 20:43
i'll merge it
thank you :)
dalek ast: 2274ac4 | Nami-Doc++ | S03-operators/orelse.t:
Swap `try`s for `lives_ok`
ast: d90866d | Nami-Doc++ | S03-operators/orelse.t:
Swap `$result` variables for `try`-`die`-`CATCH`
ast: 57c7c92 | timo++ | S03-operators/orelse.t:
Merge pull request #39 from Nami-Doc/patch-1

Swap `try`s for `lives_ok`
vendethiel thanks ! :)
timotimo no, thank you! :P
lizmat runs a new spectest
20:44 Ven left
lizmat not ok 9 - flunk orelse shouldn't rethrow exceptions# TODO orelse exception semantics 20:49
oh noes!
in block at t/spec/S03-operators/orelse.rakudo.moar:17
in block at t/spec/S03-operators/orelse.rakudo.moar:15
also: why isn't there a plan in that test-file ?
plans are good!
lizmat hopes timotimo watches 20:50
timotimo fair enough 20:51
so that test file aborts there
vendethiel: we need to cause the exception to be considered "handled"
20:51 ajr_ left
lizmat isn't that just an attribute you can set on the Exception object ? 20:52
vendethiel timotimo: wdym ?
shouldn't get to `CATCH` if `orelse` works, right ? and there's the try ?
timotimo i need to look closely again :) 20:54
i was watching a sonic 3 speedrun on the side ~_~ 20:55
vendethiel oooh, a speedrun lover :D.
timotimo if CATCH doesn't have a default { } we have to manually set $!.handled
20:55 flussence joined
vendethiel then just adding default should do, right ? 20:55
r: try { die "rip"; } 20:57
camelia ( no output )
vendethiel r: try { die "rip"; CATCH {} }
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«rip␤ in block at /tmp/tmpfile:1␤␤»
vendethiel alrighty.
r: try { die "rip"; CATCH { default {} } }
camelia ( no output )
timotimo do you have a commit bit to the roast repo?
i'd be fine with you applying that fix directly
vendethiel I don't
timotimo i don't have permissions to give, i believe 20:59
vendethiel PR is fine too 21:00
21:00 Rotwang left 21:01 rurban joined
timotimo for now :) 21:01
vendethiel that certainly is motivating.
"whoaw, I'm not only annoying people on IRC now !"
Ask I shall try not to fail. github.com/perl6/roast/pull/40 21:03
dalek ast: 5493dc6 | Nami-Doc++ | S03-operators/orelse.t:
Use `default` to mark exceptions as handled
 21:04
ast: 9191138 | timo++ | S03-operators/orelse.t:
Merge pull request #40 from Nami-Doc/patch-2

Use `default` to mark exceptions as handled (instead of skipping the rest of the test fiile)
21:05 rurban left 21:07 pdcawley left 21:16 pippo joined
pippo Hello #perl6 21:16
21:17 spider-mario left
pippo p6: say "[aaaa]" ~~ / '[' ~ ']' .+/; 21:17
camelia niecza v24-109-g48a8de3: OUTPUT«Unhandled exception: Unable to parse anonymous regex␤Couldn't find final ']'; gave up␤ at /home/p6eval/niecza/lib/CORE.setting line 1536 (die @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2962 (Cursor.FAILGOAL @ 6) ␤ at /tmp/tmpfile l…»
..rakudo-jvm c2cc4f: OUTPUT«Unable to parse expression in ; couldn't find final ']' ␤ in any FAILGOAL at gen/jvm/stage2/QRegex.nqp:1146␤ in method ACCEPTS at gen/jvm/CORE.setting:12808␤ in method ACCEPTS at gen/jvm/CORE.setting:798␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-parrot c2cc4f: OUTPUT«Unable to parse expression in ; couldn't find final ']' ␤ in any FAILGOAL at gen/parrot/stage2/QRegex.nqp:1143␤ in regex at /tmp/tmpfile:1␤ in method ACCEPTS at gen/parrot/CORE.setting:12810␤ in method ACCEPTS at gen/parrot/CORE.setting:794…»
..rakudo-moar c2cc4f: OUTPUT«Unable to parse expression in ; couldn't find final ']' ␤ in any FAILGOAL at gen/moar/stage2/QRegex.nqp:1146␤ in method ACCEPTS at src/gen/m-CORE.setting:12865␤ in method ACCEPTS at src/gen/m-CORE.setting:798␤ in block at /tmp/tmpfile:1␤␤»
21:20 pippo left
FROGGS p6: say "[aaaa]" ~~ / '[' .+ ']' /; 21:20
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f, niecza v24-109-g48a8de3: OUTPUT«｢[aaaa]｣␤␤»
21:21 rindolf left 21:22 pippo joined
timotimo oh, huh? 21:22
p6: say "[aaaa]" ~~ / '[' ~ ']' (.+) /; 21:23
camelia niecza v24-109-g48a8de3: OUTPUT«Unhandled exception: Unable to parse anonymous regex␤Couldn't find final ']'; gave up␤ at /home/p6eval/niecza/lib/CORE.setting line 1536 (die @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2962 (Cursor.FAILGOAL @ 6) ␤ at /tmp/tmpfile l…»
..rakudo-jvm c2cc4f: OUTPUT«Unable to parse expression in ; couldn't find final ']' ␤ in any FAILGOAL at gen/jvm/stage2/QRegex.nqp:1146␤ in method ACCEPTS at gen/jvm/CORE.setting:12808␤ in method ACCEPTS at gen/jvm/CORE.setting:798␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-parrot c2cc4f: OUTPUT«Unable to parse expression in ; couldn't find final ']' ␤ in any FAILGOAL at gen/parrot/stage2/QRegex.nqp:1143␤ in regex at /tmp/tmpfile:1␤ in method ACCEPTS at gen/parrot/CORE.setting:12810␤ in method ACCEPTS at gen/parrot/CORE.setting:794…»
..rakudo-moar c2cc4f: OUTPUT«Unable to parse expression in ; couldn't find final ']' ␤ in any FAILGOAL at gen/moar/stage2/QRegex.nqp:1146␤ in method ACCEPTS at src/gen/m-CORE.setting:12865␤ in method ACCEPTS at src/gen/m-CORE.setting:798␤ in block at /tmp/tmpfile:1␤␤»
FROGGS timotimo: that is known
timotimo oh?
FROGGS that form does notback of properly
timotimo ah
FROGGS s/not/not /
timotimo it eats the ']' as a .
jnthn I don't think the ~ form is meant to
pippo "[a]" ~~ /'[' ~ ']' . /; 21:24
FROGGS m: "[a]" ~~ /'[' ~ ']' . /;
camelia ( no output )
FROGGS m: say "[a]" ~~ /'[' ~ ']' . /;
camelia rakudo-moar c2cc4f: OUTPUT«｢[a]｣␤␤»
FROGGS m: say "[a]" ~~ /'[' ~ ']' .+ /;
camelia rakudo-moar c2cc4f: OUTPUT«Unable to parse expression in ; couldn't find final ']' ␤ in any FAILGOAL at gen/moar/stage2/QRegex.nqp:1146␤ in method ACCEPTS at src/gen/m-CORE.setting:12865␤ in method ACCEPTS at src/gen/m-CORE.setting:798␤ in block at /tmp/N4pOHKF1ZV:1␤␤»
lue I think r-m has far more faith in my computer than most; it was 14 threads when I bothered checking htop how part of my program was going :)
timotimo m: say "[aaaaa]" ~~ / '[' ~ ']' .+?
camelia rakudo-moar c2cc4f: OUTPUT«===SORRY!=== Error while compiling /tmp/WZRmPuNj5y␤Regex not terminated␤at /tmp/WZRmPuNj5y:1␤------> say "[aaaaa]" ~~ / '[' ~ ']' .+?⏏<EOL>␤ expecting any of:␤ postfix␤ infix stopper␤ …»
timotimo m: say "[aaaaa]" ~~ / '[' ~ ']' .+? /
camelia rakudo-moar c2cc4f: OUTPUT«Unable to parse expression in ; couldn't find final ']' ␤ in any FAILGOAL at gen/moar/stage2/QRegex.nqp:1146␤ in method ACCEPTS at src/gen/m-CORE.setting:12865␤ in method ACCEPTS at src/gen/m-CORE.setting:798␤ in block at /tmp/6Y2sPHwzvT:1␤␤»
timotimo even with a frugal quantifier? 21:25
21:26 bjz left, bjz joined
FROGGS TimToady++ commented on that once, and IIRC it is as jnthn++ says 21:26
timotimo oh hm 21:27
fair enough
FROGGS definitely a WAT though :o) 21:28
timotimo you know what we can do?
jnthn Well, it's designed to help you get good errors when you're *parsing*.
timotimo we could warn if the regex between the [ and ] is nonspecific enough to eat up the ] 21:29
and then if we find any "]" exactly one step before the current cursor position
gtodd I like how rakudo-parrot , rakudo-jvm , rakudo-moar all talk to me when I send code strings to rakudo ... but what is "c2cc4f:" ? 21:30
timotimo we add a little help text to the error
gtodd: it's the commit id of rakudo that was used to compile the evalbot's current revision
gtodd oh perfect
timotimo gtodd: if you're interested in doing a little perl5 work, you could implement a function to shorten that part of the message for us
FROGGS gtodd: see:
t;dalek> rakudo/nom: c2cc4fd | (Timo Paulssen)++ | t/spectest.data:
run the new orelse tests
gtodd I thought it was some strange unicode my irc client was munging 21:31
lizmat runs a spectest again
21:33 Mouq joined
lue I don't suppose there's a $*MAX_THREADS or something where I can say "My computer can only safely do 3 threads here", e.g. sub intense_func() { $*MAX_THREADS = 3; ... } ? 21:34
Mouq r: say "abcd[efgh], something[bananas[what, is_up]]" ~~ ms/(\w+[ \[ ~ \] <sub-expr=~~> ]?) *% \, /
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«｢abcd[efgh], something[bananas[what, is_up]]｣␤ 0 => ｢abcd[efgh]｣␤ sub-expr => ｢efgh｣␤ 0 => ｢efgh｣␤ 0 => ｢something[bananas[what, is_up]]｣␤ sub-expr => ｢bananas[what, is_up]｣␤ …»
timotimo there is, lue 21:35
you can create your own $*SCHEDULER
and everything will use that by default
21:35 kurahaupo left
timotimo the ThreadPoolScheduler has a value you can set for the max number of threads 21:35
lizmat timotimo: looks fine here new
timotimo lizmat: excellent. lizmat++ vendethiel++
vendethiel my first karma \o/ 21:36
timotimo i'm convinced it won't be your last :)
lue timotimo: OK, 'cos while running ≈14 subprocesses of perl6-m is cool, it makes my computer stutter a bit, and it would likely be way more painful for perl6-j :)
timotimo it may be interesting to check out an environment variable for the construction of the "default" scheduler. jnthn? 21:37
gtodd perl6-m perl6-j ... does that mean there's a perl6-r ?
Mouq gtodd: There's a perl6-p
lizmat m: $*SCHEDULER = ThreadPoolScheduler.new( max_threads = 4 )
camelia rakudo-moar c2cc4f: OUTPUT«===SORRY!=== Error while compiling /tmp/VuRqN0xoB9␤Preceding context expects a term, but found infix = instead␤at /tmp/VuRqN0xoB9:1␤------> = ThreadPoolScheduler.new( max_threads =⏏ 4 )␤»
gtodd Mouq: p for parrot ...
Mouq gtodd: Yup
lizmat m: $*SCHEDULER = ThreadPoolScheduler.new( max_threads => 4 ) 21:38
camelia ( no output )
lue I don't suppose there's a way to say "number of cores plus 1" easily, then?
(maybe I could give my script a -j option and let the user take care of it, like make does :P)
lizmat the code says ":$!max_threads = 16", patches welcome :-) 21:39
Mouq gtodd: Note that Parrot, JVM, and MoarVM are all run on the same front-end (Rakudo Perl 6), and that we often build Rakudo for multiple backends simultaneously, since the different backends are good for different things
timotimo i thought it was already doing that, but it turns out it's set to 16 by default
that seems a bit high to me
lue that would explain the ≈14 threads I saw earlier :)
Mouq So we install to perl6-{m,j,p} to allow for this 21:40
timotimo (and whatever comes first in Configure.pl's --backends flag gets to be "perl6")
lizmat I guess we could use an environment variable here ?
RAKUDO_MAX_THREADS=4 perl6-m foo 21:41
anyone?
timotimo +1 from my side
lue +1
jnthn Note that the number of threads you want is normally a property of system load, program behavior, etc. 21:42
lue though I wouldn't mind if Perl 6 had a commandline option too/instead...
jnthn It won't be getting one. I'm reluctant to give it an env var.
gtodd hmm so if I want to look up docs on compile/run time constants and variables (the equivalent of perldoc perlvars in pumpkin-perl-5) I type "p6doc variables" and ... well I get what I want but I don't know where I am in the documentation ... I seem to be inside a terminal pager .... 21:43
lue jnthn: yeah, I was doing fine with a 16 thread default until the part that's a bit more computationally expensive :)
jnthn lue: Right, and 4 is too few if you're doing lots of I/O
lue: The scheduler needs to learn to dynamically work out a sane number of threads 21:44
Mouq gtodd: Yeah… You probably want to look at S28, but that doesn't necessarily match the implementation
jnthn lue: By observing that adding more than a certain number makes things no better.
Mouq gtodd: Note that p6doc also has a web setup: doc.perl6.org/
gtodd .. (grepping in ps auxww tells me I'm in /home/gtodd/perl6/doc/bin/../lib/variables.pod) ... the pods don't seem as "structured" as p5 21:45
Mouq: yeah
lue jnthn: In my case, I'm calling perl6-m subprocesses to process Pod6 docs, which is where the "oh, I'm doing a lot of stuff right now" feeling occurred on a 16 thread maximum.
Mouq synopsebot: ping? S03 21:46
lue synopsebot only triggers on syn:line
Mouq Oh :P
gtodd Mouq: just noting there seems to be none of the structured documentation cues one normally sees in a pod ... yet ... but I guess POD formats have changed for p6 and are not yet complete 21:47
dalek ecs: f136000 | (Elizabeth Mattijsen)++ | S17-concurrency.pod:
Add "Redeemed" status and .redeem to Promise

This would be the equivalent of Channel.close.
lue gtodd: what do you mean?
Mouq gtodd: Could you be more specific? What's unstructured about what we have? I'd like to improve what we have if possible
jnthn lizmat: I...doubt that's going to last.
Mouq has never used p6doc, note 21:48
21:48 WJB left
lizmat jnthn: why? 21:48
Mouq So many questions
:)
jnthn lizmat: Well, for one because a promise being kept/broken means it is implicitly "closed" to further values. 21:49
lizmat But I would like a way to mark the Promise as "the value of the Promise has been processed, don't call me again"
s/value/result 21:50
Mouq There is a lot of work we need to do with the documentation
timotimo so basically a back-channel?
Mouq I suppose I'll start planning out the steps for implementing smls++'s ideas
lue gtodd: what about the docs are unstructured to you, out of curiosity?
lizmat the problem with the current Promise is, that you can fetch its .result as many times as you want
timotimo Mouq: these ideas were mostly the style and layout of the documentation website? 21:51
Mouq timotimo: They implied a cleaner overall structure though
timotimo: And this structure could be back-ported to p6doc
lizmat m: my $p = Promise.new; $p.keep(42); say $p.result; say $p.result 21:52
camelia rakudo-moar c2cc4f: OUTPUT«42␤42␤»
jnthn lizmat: But...what if multiple things are processing it?
lizmat well, then don/t .redeem it
21:52 WJB joined
lizmat use the vow, and only through the .vow, redeem it 21:52
21:52 pippo left
lizmat when you know that all multiple things have processed it 21:53
timotimo i don't see the sense i nthat :\
jnthn Now the design problems with winner/outcome/wahtever have polluted the well designed promise.
lizmat it would make winner { } make sens
timotimo i thought the redemption was the way to signalize completion of processing backwards to the "owner" of the promise
jnthn lizmat: No, it will just create horrible action that a distance that we'll end up regretting for years aftewrwards. 21:54
Maybe winner should just be about channels, if it's going to make us ruin Promise like this.
lue gtodd: are you saying that the files in p6doc aren't organized, or that there aren't consistent sectionings, or...? 21:55
jnthn And we introduce a Promise.Channel coercion if you really want a channel out of a Promise to use with winner.
lizmat jnthn: I'm not seeing the action at a distance problem, if you could only redeem through the vow
timotimo lizmat: if you can only redeem promises through the vow, how do you get the information that processing has finished from the consumers? 21:57
lizmat well, then maybe you cannot redeem promises that have been vowed ?
vendethiel m: my $p = Promise.new; $.keep({ say "hay"; 42 }); say $p.result; say $p.result;
camelia rakudo-moar c2cc4f: OUTPUT«===SORRY!=== Error while compiling /tmp/lsfOB0Q7Ok␤Variable $.keep used where no 'self' is available␤at /tmp/lsfOB0Q7Ok:1␤------> = Promise.new; $.keep({ say "hay"; 42 })⏏; say $p.result; say $p.result;␤»
timotimo the only thing i could imagine is Promise.care giving a handler that you can call .forget on and when the last .care has been .forget'd it would redeem ti, but that seems fragile and racy 21:58
vendethiel m: my $p = Promise.new; $p.keep({ say "hay"; 42 }); say $p.result; say $p.result;
camelia rakudo-moar c2cc4f: OUTPUT«Block.new()␤Block.new()␤»
timotimo m: my $p = Promise.new; $p.keep({ say "hay"; 42 }); say $p.result.(); say $p.result.();
camelia rakudo-moar c2cc4f: OUTPUT«hay␤42␤hay␤42␤»
vendethiel that's to be expected tho. 21:59
timotimo what exactly?
jnthn lizmat: vow is used by (hopefully) all the internal promise constructors to make sure only they can keep/break.
lizmat what I was getting to, is a sane behaviour of: 22:00
loop { winner { done @p { ... } } }
where promised that are redeemed, will never match a "done"
*promises
vendethiel timotimo: the double "hay" 22:01
lizmat in other words, a "Redeemed" status would make winner {} sane
22:01 a3gis joined
timotimo vendethiel: what was your code trying to prove? o_O 22:02
22:02 rurban joined
vendethiel timotimo: I wanted to know if .result would get the block called 22:02
timotimo ah
vendethiel and if so, whether it'd be called once or twicve
timotimo now we know :)
lizmat and in my proposal, if you would do $p.redeem between the two calls 22:03
the second one would be a noop now
timotimo yeah, at the moment you'd need to kick them out of the list 22:04
lizmat which is silly bookkeeping
vendethiel r: try { die "abc"; CATCH { when Str { say "hey" } } } 22:05
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«abc␤ in block at /tmp/tmpfile:1␤␤»
lizmat anyways, I'm almost done with speccing "outcome"
jnthn: if you don't like that, we just revert :-) 22:06
timotimo one way i could imagine this to go is having a lock and calling @promises>>.then({ $lock.protect: { do something } })
vendethiel r: say "abc" ~~ Str
camelia rakudo-parrot c2cc4f, rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«True␤»
vendethiel Not sure why I can't `catch when str` ?
timotimo that'll do something similar to the winner loop
lizmat timotimo: locks are evil and shouldn't be visible like that
timotimo vendethiel: because die will wrap a string up into an X::AdHoc
r: try die "oh god what?"; say $!.DUMP
camelia rakudo-parrot c2cc4f: OUTPUT«X::AdHoc<1>(␤ :payload(▶"oh god what?"),␤ :$!ex(Exception<3>(...))␤)␤»
..rakudo-jvm c2cc4f, rakudo-moar c2cc4f: OUTPUT«X::AdHoc<1>(␤ :payload(▶"oh god what?"),␤ :$!ex(BOOTException<3>(...))␤)␤»
22:07 rurban left
timotimo lizmat: well, in order to serialize stuff ... hmm. 22:07
well, i suppose you can also .>>then({$my_channel.send($stuff)}) and serialize it that way
and then for @$my_chanel { ... }
but those are all hacks i'm afraid
dalek c: 8b066e3 | Mouq++ | lib/Bag.pod:
`==` → `===` in Bag.pod

  `==` is numeric comparison. `===` does what we want here.		 22:08
22:13 kivutar left
dalek ecs: a332fbc | (Elizabeth Mattijsen)++ | S17-concurrency.pod:
Spec "outcome/whenever" instead of "winner/done/more"

For discussion, and for a test implementation. Please comment / fire on it / shoot it down if you think you should. By no means do I feel this is the final thing. It's just that winner {} currently is *so* broken, we either need to fix that, or come up with something else. This is just my something else :-)		 22:19
lizmat hmmm.. maybe done on a Promise should allow you to redeem the promise 22:21
done $p1 { $_.redeem }
lue Is it normal for panda install on perl6-m to die with Nil\nNo such method 'get-project' for invocant of type 'Any' ? 22:22
lizmat perhaps .redeem should be a separate attribute on the Promise 22:23
simply to mark it as "handled" as it were
timotimo it could apply a role perhaps 22:24
lizmat mayhap 22:27
22:28 kurahaupo joined
vendethiel lizmat++ # changing "winner" keyword 22:29
22:32 kurahaupo left 22:34 araujo joined
lizmat gnight #perl6! 22:36
jnthn 'night, lizmat 22:37
Mouq night lizmat!
22:38 kurahaupo joined
timotimo gnite liz 22:39
22:45 kivutar joined
timotimo r: sub tester(*@args, *%kw) { say (:@args); say (:%kw) }; tester(1, 2, 3) :{ say "hi"; } 22:50
camelia rakudo-moar c2cc4f: OUTPUT«===SORRY!===␤Cannot find method 'has_compile_time_value'␤»
..rakudo-parrot c2cc4f, rakudo-jvm c2cc4f: OUTPUT«===SORRY!===␤No such method 'has_compile_time_value' for invocant of type 'NQPMu'␤»
timotimo so allegedly that is supposed to work? o_O
22:56 kurahaupo left 23:01 colomon joined 23:03 rurban joined 23:05 dmol left 23:08 rurban left
vendethiel timotimo: as an adverb? 23:13
23:13 colomon left
timotimo This just naturally falls out from the preceding rules because the adverbial block is in operator position, so it modifies the "find operator". (Parens aren't considered an operator.) 23:13
find($directory) :{ when not /^\./ }
vendethiel uhhh 23:14
23:24 kurahaupo joined
Mouq doesn't understand 23:24
r: sub tester(*@args, *%kw) { say (:@args); say (:%kw) }; tester(1, 2, 3) :( 1 )
camelia rakudo-moar c2cc4f: OUTPUT«===SORRY!===␤Cannot find method 'has_compile_time_value'␤»
..rakudo-parrot c2cc4f, rakudo-jvm c2cc4f: OUTPUT«===SORRY!===␤No such method 'has_compile_time_value' for invocant of type 'NQPMu'␤»
Mouq What is that supposed to do again?
What is the "find operator"? 23:25
std: sub tester(*@args, *%kw) { say (:@args); say (:%kw) }; tester(1, 2, 3) :( 1 )
camelia std e347792: OUTPUT«ok 00:01 133m␤»
Mouq n: sub tester(*@args, *%kw) { say (:@args); say (:%kw) }; tester(1, 2, 3) :( 1 )
camelia niecza v24-109-g48a8de3: OUTPUT«===SORRY!===␤␤Action method fakesignature not yet implemented at /tmp/bNv5OYe3eo line 1:␤------> gs); say (:%kw) }; tester(1, 2, 3) :( 1 ⏏)␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niec…»
Mouq n: sub tester(*@args, *%kw) { say (:@args); say (:%kw) }; tester(1, 2, 3) :{ True }
camelia niecza v24-109-g48a8de3: OUTPUT«"args" => (1, 2, 3).list␤"kw" => {"" => { ... }}.hash␤»
Mouq n: sub tester(*@args, *%kw) { say (:@args); say (:%kw) }; tester(1, 2, 3) :{ say "hi"; } 23:26
camelia niecza v24-109-g48a8de3: OUTPUT«"args" => (1, 2, 3).list␤"kw" => {"" => { ... }}.hash␤»
23:29 BenGoldberg joined
Mouq std parses tester(1, 2, 3) and :{ True } as two statements in the same statementlist… 23:30
23:30 btyler joined 23:35 colomon joined 23:44 denis_boyun___ left 23:47 colomon left 23:49 kivutar left 23:54 colomon joined 23:56 a3gis left
timotimo Mouq: i think it's in " " because it's actually not a real "operator" per se; it's just a regular sub that gets postcircumfix:<( )>'d 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!