»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:13 molaf_ joined 00:17 molaf left 00:19 donri joined 00:28 tokuhirom joined 00:31 icwiener left 00:32 replore joined, thou left 00:35 thou joined
jnthn 'night, #perl6 00:43
japhb phenny, ask sorear You mentioned wanting to do a Niecza/C. Does that mean "implemented in C, targeting some VM or JIT" or "implemented in C, targeting C as an intermediate language", or some other variation entirely? 00:44
phenny japhb: I'll pass that on when sorear is around.
japhb 'night, jnthn
colomon japhb: sorear was talking about what GC to use, so I'm pretty sure he means targeting C as an intermediate language. 00:46
diakopter no, I'm pretty sure he means writing his own VM/interpreter/JIT in C. 00:47
colomon sure, could be that too. 00:48
japhb colomon, in that case I wonder if C-- might be a better intermediate language choice. He does know GHC after all, so he at least has seen the cmm bits there. :-)
diakopter, *chuckle* And now I'm back to knowing nothing of his plans again, with two of you offering opposite interpretations. ;-) 00:49
diakopter I'm nearly certain he doesn't want to rely on a C compiler at runtime 00:51
00:54 hundskatt left 00:57 tokuhirom left
colomon diakopter: going back to the perl 5 model, eh? that certainly fits with what he said, and seems to make sense. 00:58
diakopter or parrot model
01:02 cooper left 01:03 lestrrat left
colomon I'm just guessing, mind you, but I'd be really surprised if he were thinking the parrot model (ie, a new universal VM loosely coupled to niecza) rather than the p5 model (custom VM designed to do exactly what niecza needs). 01:03
01:04 lestrrat joined
japhb colomon, perhaps diakopter meant the parrot model in the sense of lower-level, byte-code and/or jit-based rather than just interpreting the optree as perl5 does. 01:06
colomon ah, sure, that could make sense, too. :) 01:07
01:22 plutoid joined 01:27 thou left 01:50 woosley joined 02:16 hypolin joined 02:26 donri left
dalek kudo/nom: 652952a | colomon++ | README:
Add --gen-nqp to the Configure instructions.
02:30
02:38 envi_ left 02:44 Guest28440 left 02:56 wolfman2000 joined 02:57 agentzh joined 03:02 satyavvd joined 03:06 lestrrat left 03:07 lestrrat joined
colomon Where does the updated Pugs repo live now that the revival effort is underway? 03:24
So, here's my first stab at the Advent Day 1 post. Comments and suggestions would be very welcome. perl6advent.wordpress.com/?p=735&am...eview=true 03:27
(You may need Advent editing permission to see that preview...)
03:42 worr left 03:47 preflex left 03:49 preflex joined
sorear good * #perl6 04:08
phenny sorear: 00:44Z <japhb> ask sorear You mentioned wanting to do a Niecza/C. Does that mean "implemented in C, targeting some VM or JIT" or "implemented in C, targeting C as an intermediate language", or some other variation entirely?
sorear japhb: I am _not_ planning to generate C 04:12
I'm going for more of a threaded code interpreter like what Perl 5 does but with better dcache behavior 04:13
04:41 agentzh left 04:55 ethndbst left 05:43 ethndbst joined, ethndbst left, ethndbst joined
dalek ecza: 0b17f48 | sorear++ | docs/release.txt:
Add notes on how to cut a Niecza release
05:52
ecza: eb42dfa | sorear++ | FETCH_URL:
Update FETCH_URL
06:01 localhost left 06:03 localhost joined 06:15 kaleem joined 06:25 ruz left 06:26 clkao left 06:32 molaf_ left 06:38 wolfman2000 left 06:42 rsimoes1 is now known as rsimoes 06:44 Chillance joined 06:50 koban joined 07:04 fridim_ joined
moritz good morning 07:07
sorear good morning moritz
moritz tomorrow is the first of December
if we want a Perl 6 advent calendar, soembody needs to write a first post 07:08
and people need to commit to dates and topics
07:19 replore left
Woodi I would like to read few "perlish" topics like about debugging/profiling tools (in Parrot too) or Grammar::Trace (me seen video) 07:21
moritz Woodi: then do. Add your name and topic(s) to misc/perl6advent-2011/schedule in the perl6/mu repository 07:25
erm
sorry, I thought you wanted to write about them
masak morning, #perl6
moritz we have no lack of topics, we have a lack of people committed to dates and topics
good morning masak 07:26
masak if we don't have enough people committed to dates and topics by the end of the day, maybe it's better to not run an advent calendar this year? 07:27
finding people to write the next day's post tends to be the tricky part (if not the most time-consuming, which is probably writing the post) 07:28
moritz well, quite a few people seemed to be willing to do something
it would be a shame not to do
but I get your point
I wonder what's better, an advent calendar with only entries for half the days, or none at all 07:29
anyway, my current real-life situation does not allow me to spend as much time on it as last year (where I wrote several last-minute posts to fill gaps) 07:32
masak same here. 07:33
I'm going on a vacation with at best reduced internet access for two weeks, starting Friday. 07:34
07:37 hundskatt joined 07:38 pmurias joined, ethndbst left
pmurias sorear: why write an interpreter instead of generating C/llvm ir? 07:39
07:39 preflex_ joined 07:40 preflex left, preflex_ is now known as preflex
sorear pmurias: llvm is SLOW 07:40
pmurias: and gcc is slower
using llvm would erase all the wins from abandoning Mono
o/ masak
pmurias sorear: what would the wins of abandoning Mono be? 07:41
sorear faster startup, much more portability, less memory use
07:42 baest left
sorear moritz: I'm not sure I get it. Didn't we cover all the topics last year? 07:43
What's left?
07:44 baest joined
pmurias sorear: you could try targeting the V8 IR 07:46
masak sorear: I'm sure the set of Perl 6 topics has a cardinality bigger than 50 :) 07:47
pmurias sorear: re gcc is slow, yes it used to be a problem with mildew 07:48
moritz sorear: there are infinitely many topics. See topic-brainstorming in perl6/mu/misc/perl6advent-2011 07:52
07:53 obra joined, alvis left
jnthn moritz: Gotta commute/do onsite $dayjob in a few minutes...I can commit to 2 or maybe 3 posts, including one on Grammar::Debugger and one on meta-programming. 07:55
07:56 obra left
sorear wonders what the old pumpking dropped in for 08:05
moritz: ah, cool 08:06
sorear can't do "a look at Niecza" without first getting a better idea of what Niecza is and why it still exists 08:07
incidentally, the unicode folks made the clarification I wanted to tn18 before I had a chance to ask for it
08:08 arlinius left
masak "the old pumpking" sounds like something out of a children's story :) 08:08
08:08 alvis joined
sorear masak: until last week or so, obra was the p5 pumpking (now it's rjbs) 08:09
masak nod 08:14
oh, the switch wasn't until last week? I'd have thought it happened a month ago or so.
sorear maybe, my sense of time ain't great 08:15
sorear sleep 08:16
08:19 am0c^ joined 08:29 am0c^ left
moritz phenny: ask sorear if he'd like to write something for p6advent about his Unicode work for niecza 08:31
phenny moritz: I'll pass that on when sorear is around.
moritz jnthn: that's great. Could you please chose slots for that in github.com/perl6/mu/blob/master/mi...1/schedule ? 08:33
moritz just discovered perldoc.perl.org/pstruct.html 08:35
jnthn: btw I ran test-summary.pl yesterday, 17690 passes 08:39
08:39 mj41 joined
jnthn moritz: najs! 08:39
er. nice!
moritz I've got 6 failures in t/spec/S05-mass/properties-general.t after [Coke]++ unfudged a bit. I guess they depend on the ICU version 08:42
oh, and whiteknight++ pondered bundling icu with parrot again yesterday
08:43 stepnem left 08:49 stepnem joined
jnthn moritz: ooh, progress :) 08:51
moritz: I can't get at the mu repo from this $dayjob place...will add myself this evening. 08:52
moritz jnthn: ok. Or tell me the dates, and I'll enter you 08:54
08:54 am0c^ joined
moritz erm, it 08:55
whatever
jnthn lol
moritz: not 7th-9th and not 15th or 16th. Other than that, any days are fine. 08:56
08:58 arlinius joined
dalek : 7c49eae | moritz++ | misc/perl6advent-2011/schedule:
sign up jnthn++ for day 2 and 14
08:59
masak jnthn++
I might have time to prepare-write something tomorrow, if I have time left over after blogging. 09:00
but I don't want to sign up until I have something prepared.
I could sign up already for any date after the 18th.
moritz masak: please do
masak haven't thought about what to write about yet, but if signing up my name alone is OK, I'll do that. 09:01
moritz it is
masak does so, then
09:03 koban left
dalek : c8126c4 | (Carl Mäsak)++ | misc/perl6advent-2011/schedule:
Signed up masak for 19th and 22nd
09:03
moritz masak++
09:04 domidumont left, am0c^ left 09:06 domidumont joined
snarkyboojum abandoning Mono might forgo all the library goodness that Colomon was praising the other day too tho :( 09:08
moritz kinda thinks there are many good C libraries too 09:09
snarkyboojum no doubt, though using C libraries is arguably more involved that a "platform" like Mono
I played around porting dev.aboutus.org/2011/07/03/getting-...rsing.html to Perl 6 which runs better on nom now - might make an interesting Advent post 09:11
though there might already be grammar'y type posts in there 09:12
Perl 6 naturally fits that kind of "exploratory parsing" niche, without having to install software like peg/leg (which is very good btw) 09:14
or not! 09:18
09:18 mkramer joined
snarkyboojum i.e. one of the simple examples is using more than 3GB of memory atm :( 09:19
but at least it hasn't coredumped yet :)
09:20 mkramer left 09:21 mkramer joined, mkramer left, mkramer joined 09:22 woosley left
pmurias snarkyboojum: re abandoning Mono, is the plan to abandon it or merely add an additional backend? 09:32
masak does anyone know of any programming language besides Java and Perl 6 where enums are reference types?
snarkyboojum pmurias: no idea - just reading sorears earlier comments 09:33
masak snarkyboojum: I think you should definitely go for an "exploratory parsing" post. for one, because I'd like to read it :) but also because I know what you mean and I agree that it's one of the strengths of Perl 6 in general and of Niecza/Rakudo in particular. 09:34
09:35 mkramer left
moritz snarkyboojum: I'd very much welcome such a post 09:35
09:35 mkramer joined, mkramer left
pmurias masak: enums are mutable in java? 09:36
moritz snarkyboojum: grammars are a very central p6 feature (IMHO), and we can't show it off too much :-)
snarkyboojum it seems it's the .perl on the match object in my little script that is the culprit for making the thing effectively hang :(
moritz masak: cool :)
moritz snarkyboojum: on rakudo, try .gist for inspecting 09:37
09:37 dakkar joined
snarkyboojum moritz: good idea - will give that a go 09:37
moritz snarkyboojum: but please submit a bug report too, I'd like to fix .perl eventually
masak pmurias: essentially yes, but it's considered poor taste to give enum values mutable state.
pmurias: or, as it was aptly put on stackoverflow "yes, but that would be *evil*" 09:38
09:38 mkramer joined
moritz (nearly) everything is mutable, if you go to a sufficiently low level :-) 09:38
snarkyboojum I am trying to dump 16814 "facts" which have been parsed out of a text file (via .perl or .gist - both of which are v. slow)
moritz erm, yes. I can imagine that 09:39
snarkyboojum if just just count them, all is good, so it's string performance again I suppose
moritz or maybe iteration preformance 09:41
snarkyboojum ah
moritz note that .perl is optimized for correctness, not for speed or readability
snarkyboojum: maybe I add you to the p6advent schedule? if yes, which day? 09:42
snarkyboojum moritz: ok - stick me in there in the middle somewhere :) 09:43
09:43 mkramer left
moritz snarkyboojum: 12th then? :-) 09:44
snarkyboojum major project at work goes "live" on Dec 14, so around there ;)
moritz it's as middle as it gets :-)
around there, or later?
09:45 mkramer joined
snarkyboojum 12 would be probably be a great excuse for procrastination 09:45
moritz ok, 12 it is then
snarkyboojum "lock it in eddy"
dalek : 7b043af | moritz++ | misc/perl6advent-2011/schedule:
[p6advent] snarkyboojum++ volunteers for day 12
09:46
: 0005bed | moritz++ | misc/perl6advent-2011/ (2 files):
[p6advent] sign me up for day 3, Buf and binary IO
09:50
09:53 domidumont left 09:54 envi_ joined 10:05 hundskatt left 10:06 hypolin left
moritz nom: say '' ~ Buf.new 10:07
p6eval nom 652952: OUTPUT«Buf<656769015169214330>␤»
10:07 mkramer left 10:08 mkramer joined
masak ooh, I know! I'll write about operator precedence! I don't think anyone's done that before. 10:19
and it's a topic I like ^_^
moritz nom: gist.github.com/1408580 10:22
p6eval nom 652952: OUTPUT«møþ␤møþ␤»
moritz it seems that $*OUT.write(Buf) is broken :(
10:24 daxim joined
jnthn Just $*OUT? 10:24
moritz hm, no idea
moritz checks
nope, other :w file handles too 10:26
hm
oh
I need to open it as :bin
10:27 mkramer left
moritz nom: nqp::say(nqp::typeof__SP(pir::getstdout__P())) 10:28
p6eval nom 652952: OUTPUT«===SORRY!===␤Unrecognized nqp:: opcode 'nqp::typeof__SP' at line 1, near ")"␤»
moritz nom: nqp::say(pir::typeof__SP(pir::getstdout__P()))
p6eval nom 652952: OUTPUT«FileHandle␤»
moritz nom: nqp::say(pir::getstdout__P().encoding) 10:29
p6eval nom 652952: OUTPUT«utf8␤»
10:31 satyavvd left
moritz should .write automatically temporarily switch to binary IO? or die? 10:34
doing something non-sensical semems wrong to me :-)
moritz goes with DWIM for now
10:37 pmurias left
masak die would be the conservative option, but switching to binary has a certain allure, too, I guess. 10:38
moritz nom: $*OUT.write($*IN.read(200)) 10:39
p6eval nom 652952: OUTPUT«Land der Berge, Land am Strome,␤Land der Ãcker, Land der Dome,␤Land der Hämmer, zukunftsreich!␤Heimat bist du groÃer Söhne,␤Volk, begnadet für das Schöne,␤vielgerühmtes Ãsterreich,␤vielgerühm»
moritz ah, same problem 10:40
masak .write should always work on Buf, no? 10:41
moritz and .read returns one, yes 10:42
masak oh, but the problem is $*OUT being set to binary or not? I see. 10:43
moritz curiously I think that $*IN doesn't need to be set to binary already
10:44 domidumont joined 10:45 agentzh joined 10:46 plutoid left
dalek kudo/nom: 80f6ce0 | moritz++ | src/core/IO.pm:
make IO.write DWIM if the handle is not binary yet
10:50
moritz erm 10:53
does anybody else find the section on Blob literals confusing? 10:54
:16{A705E} a blob4, 4 bits per column
colomon masak: what can you tell me about the pugs revival effort (for the Day 1 post)?
moritz so, to build a blob8, you need to write :256{...}? And what do you put in there? 10:55
colomon: ingy fixed some build issues, apart from that nothing has happened, only planning 10:56
(re pugs)
colomon where's the repo for it? is it still audreyt's, or is there a fork somewhere?
moritz github.com/perl6/Pugs.hs 10:57
std: :2{0010} 10:58
p6eval std be1f10e: OUTPUT«===SORRY!===␤Malformed radix number at /tmp/vOx4dPvMHA line 1:␤------> :2⏏{0010}␤ expecting any of:␤ statement end␤ statement list␤Parse failed␤FAILED 00:01 118m␤»
10:58 mkramer joined
colomon moritz++ 11:04
masak colomon: an accurate picture of the pugs revival effort, I think, is that there's plans for it but it hasn't begun yet.
moritz I'd rather have all Blob literals generate a Buf8, and have the leading number denote the basis for it, not the column width 11:05
masak some people have gotten Pugs from a repo on their boxes. I haven't yet. haven't tried very hard, though.
moritz nom: gist.github.com/1408580 11:06
p6eval nom 652952: OUTPUT«møþ␤møþ␤»
moritz evalbot control rebuild
evalbot rebuild nom
p6eval OK (started asynchronously)
colomon Do the plans for pugs include trying to bring it closer to the current spec? 11:10
masak absolutely. 11:13
though probably more on the LHF level of things until we understand the Pugs architecture better :) 11:14
getting Pugs to run the roast tests would be a short-term goal as well. 11:15
moritz any thoughts on the Blob literal issue I raised above? 11:18
masak moritz: I thought Blob was to be an immutable Buf... 11:19
moritz yes
(I might not have made the distinction above, because rakudo only implements Buf)
but the issue is the same nonethelss 11:20
by current spec, to create a Blob8 / Buf8, you'd need to write :256{...}, and I don't know wha the '...' is supposed to be 11:21
everything else in Perl 6 that supports radix notation only goes up to 36, because we only have 10 digits and 26 ASCIIbetical characters
question to the Socket users: should .recv return a Buf? 11:25
colomon moritz: just off the top of my head (no idea what the spec says), I'd think the ... should be in hex. 11:28
anyone know if there's an easy way to get an HTML file from a github repo to actually display like a web page (instead of HTML source)? 11:30
11:33 ggoebel joined 11:38 markand joined
markand hi 11:38
daxim masak, paste.scsys.co.uk/163516
roasting is easy
markand is it still possible to use perl5 from cpan with perl6 ? 11:42
tadzik I think it's possible in niecza 11:43
11:45 whiteknight joined, whiteknight is now known as Guest622
masak daxim: nice! 11:52
daxim pugs: use perl5::Math::Trig; tan(50).say;
p6eval pugs b927740: OUTPUT«pugs: *** Unsafe function 'use' called under safe mode␤ at /tmp/Um8xTt7FyL line 1, column 1␤»
daxim markand, this does work offline.
markand okay :) 11:53
thanks
12:02 mkramer left 12:13 Psyche^ joined 12:16 Patterner left, Psyche^ is now known as Patterner 12:18 agentzh left 12:23 benabik left 12:35 hundskatt joined 12:43 mkramer1 joined 12:47 bluescreen10 joined 12:53 mtk joined
dalek kudo/nom: ff7fd2d | moritz++ | src/core/IO/Socket.pm:
.write method for sockets
12:56
moritz I've only tested that method very lightly; more testing appreciated
13:08 mtk left 13:10 benabik joined 13:12 mtk joined 13:19 tokuhirom joined 13:26 mkramer1 left, kaleem left
moritz nom: class A { has str $.x; method BUILD(:$!x) { } }; say A.new(:x<foo>).x 13:44
p6eval nom ff7fd2: OUTPUT«Cannot assign to a non-container␤ in method BUILD at /tmp/rmcjzlnZkz:1␤ in method BUILDALL at src/gen/CORE.setting:568␤ in method bless at src/gen/CORE.setting:558␤ in method new at src/gen/CORE.setting:543␤ in block <anon> at /tmp/rmcjzlnZkz:1␤ in <anon> at /tm…
moritz submits rakudobug 13:46
jnthn Hm :) 13:48
moritz nom: class A { has str $.x; method BUILD(str :$!x) { } }; say A.new(:x<foo>) 13:51
p6eval nom ff7fd2: OUTPUT«Binding to natively typed attributive parameter '$!x' not supported␤ in method BUILD at /tmp/zHH0gweEES:1␤ in method BUILDALL at src/gen/CORE.setting:568␤ in method bless at src/gen/CORE.setting:558␤ in method new at src/gen/CORE.setting:543␤ in block <anon> at /…
jnthn At least it knows it can't do that one yet :) 13:53
13:58 hundskatt left 14:04 markand left 14:14 muixirt joined
muixirt hi folks 14:16
colomon \o
muixirt does NativeCall/Zavolaj work? 14:17
moritz muixirt: there's a branch called v2 in Zavolaj that should work on very latest nom 14:18
muixirt: test very much appreciated 14:19
jnthn should merge that branch this evening. 14:23
moritz deleted the rakudo nci branch today, hoping to have remembered correctly that it was already merged 14:24
jnthn moritz: merged last night. 14:25
thanks for cleaning up after my branch mess :)
14:28 lichtkind joined
moritz how much has Zavolaj's API changed? 14:35
jnthn moritz: little 14:36
You still write "is native('library')" 14:37
moritz cool
jnthn and OpaquePointer
and so forth
moritz just the struct things is new, no?
jnthn You can use things like int8 and int16 in the parameter types
You can also specify how to marshall the Str type now
And it handles null much better
(pass a type object to pass a null)
at least, for string, struct and pointer types 14:38
yes, struct stuff is at an early stage, but works for basic cases.
muixirt jnthn: wrt zavolaj: a make install should suffice, right? get a LTA error message
jnthn muixirt: I've not tried that at all in the v2 branch; I know make test didn't work out yet 14:39
muixirt: What do you get, ooc?
muixirt nopaste.snit.ch/97385 14:41
moritz muixirt: how old/young is your rakudo build? 14:42
muixirt from today
moritz hm
and did you remember to install it?
the error message usually means that an op is not available that should be 14:43
which lets me guess that the native call ops aren't available
muixirt the nci was merged in? :-) 14:44
jnthn yes, looks like missing nqp_dyncall_ops or so
muixirt: nqp provides some native calling support. The NativeCall.pm module just puts a shiny Perl 6 wrapper around it all. 14:45
14:47 JimmyZ joined
JimmyZ good evening, #perl6 14:47
PerlJam good morning JimmyZ :)
14:47 Guest622 left
JimmyZ PerlJam: hehe :) 14:48
JimmyZ wonders how about callback
jnthn JimmyZ: I didn't get those in yet. 14:49
JimmyZ: dyncall library does provide support for that. So I hope to get to it soon.
JimmyZ jnthn: thanks ;)
benabik jnthn: Just curious: Why did you use dyncall instead of ffi in Parrot? Just being contrary or does libFFI have some variety of issue? 14:51
moritz windows build troubles
benabik dyncall looks interesting, I'll admit. :-) 14:52
jnthn benabik: Build issues with libFFI, plus it doesn't seem very actively developed. Plus I didn't want people to have to dependency hunt. But what really did it was that I wanted to get some decent 6model integration.
Now you can define a C structure in Perl 6 just by writing class and marking it "is repr('CStruct')". 14:53
benabik jnthn: Cool, cool. Looks slick.
muixirt jnthn: sorry the error message was probably caused by some stale files 14:54
14:55 jasonmay left 14:56 jasonmay joined
muixirt jnthn: why is the .so extension needed? ( is native('libfoobar.so') ) 15:08
jnthn muixirt: I think whatever you specify at the moment gets passed straight down to the OS loader. Probably should get the .so added for you. 15:12
15:12 kaleem joined
muixirt jnthn and moritz, thanks for the help and the work on nci, bigint, optimizations and all the other things! 15:17
15:23 molaf joined 15:31 ethndbst joined 15:35 MayDaniel joined 15:41 thou joined 15:52 _ilbot left, mkramer joined 15:57 mkramer left 16:00 mkramer joined 16:01 mkramer left, _ilbot joined 16:20 JimmyZ left
jnthn decommute & 16:23
[Coke] jnthn: hio. 16:25
16:29 tokuhirom left, tokuhirom joined 16:33 tokuhirom left 16:34 kaleem left
[Coke] moritz? (or anyone here familiar with autounfudge?) 16:43
I'm having trouble unfudging a file because the fudged variant is returning 1 instead of 0 for an exit code. 16:45
japhb phenny, tell moritz The spec supports :256[255,127,0,54] -- square braces, then comma separated decimal values
phenny japhb: I'll pass that on when moritz is around.
[Coke] (all tests are skipped - if you run make .../../foo.t , exit value 0. if you run the fudged variant, exit value 1 16:46
japhb moritz, My Str.Numeric rewrite already supports this.
[Coke] the fudged version has an explicit exit 1 in it.... wtf?
japhb phenny, tell moritz My Str.Numeric rewrite already supports this.
phenny japhb: I'll pass that on when moritz is around.
japhb I may have to just push that to a branch, so other people can look at it, even though it can't be merged because of the Failure handling issue ... hmmm, I'll do that next time I have access to that computer unless someone gives me a better option 16:49
16:55 kaleem joined
[Coke] wtf. made trivial whitespace change to t/spec/fudge. autounfudge begins to work. back out trivial change. still works. 16:55
*head explode* 16:56
dalek kudo/nom: f0ac58c | coke++ | tools/autounfudge.pl:
add more on-demand debug output
16:58
colomon japhb: of course! dunno how I forgot :256[255,127,0,54] , was just looking at it a few weeks ago.
PerlJam colomon: I know how you forgot it ... Perl 6 is a big language with lots of corners. :) 16:59
(fortunately most of the corners aren't sharp)
jnthn home :) 17:01
colomon PerlJam: and my brain has enormous quantities of clutter in it...
17:19 kranius left
pmichaud good morning, #perl6 17:22
jnthn hi, pmichaud
nice to see you around here again! :) 17:23
pmichaud good to be around again :)(
:)
still can't type, though. :)
PerlJam pmichaud!
pmichaud: How goes?
pmichaud things are okay here... still making many trips to doctors and the like though 17:24
17:24 kranius joined 17:25 zby_home joined
[Coke] pmichaud: o/ 17:28
17:36 clkao joined
moritz \o pmichaud 17:38
colomon hey, #phasers?
phenny moritz: 16:45Z <japhb> tell moritz The spec supports :256[255,127,0,54] -- square braces, then comma separated decimal values
moritz: 16:46Z <japhb> tell moritz My Str.Numeric rewrite already supports this.
jnthn er :) 17:39
17:39 flussence joined
moritz japhb: erm, if :256[...] returns a Blob or Buf, Str.Numeric shouldn't support it, just val(), right? 17:39
PerlJam: can I motivate you to write something for the p6advent calendar? 17:40
PerlJam moritz: I'm planning on writing something, I jsut don't know what yet.
17:41 MayDaniel left 17:47 sayu joined
PerlJam colomon++ for getting the ball rolling on this year's advent calendar 17:48
colomon PerlJam: I was really amused when I looked at the backtrace this morning and saw moritz talking about the need to figure out who was going to do day 1, when I already had the first draft of it up on the blog. :) 17:49
PerlJam colomon: Mentioning R* as a separate implementation of Perl 6 seems weird. Also, isn't there a R* release planned for Dec? Talking about the previous R* on the eve of the new R* is also a little weird. 17:50
colomon: who are "they" that have pugs building aain?
er, again
colomon masak and ingy seem to be the ringleaders.
I think that R* needs to be mentioned separately at this time. I certainly hope nom overtakes it during the month, and will edit the post accordingly when that happens. 17:51
hmmm.... how does one properly spell Masak and Ingy? Do I remember correctly that they both have umlauts? 17:54
PerlJam ingy döt net 17:55
(I don't know if he goes by that moniker "for real" though)
daxim legal name. 17:56
pmichaud I'm pretty sure it's "for real"
daxim or "wallet name" is ’nym wars parlance
colomon ah, I'll just leave the "döt net" part off. 17:57
and it's Mäsak 17:58
jnthn Yes, that's correct :) 17:59
colomon jnthn: are there lists somewhere of the 1) new features of nom (versus b) 2) current shortcomings of nom? 18:00
pmichaud ...roadmap?
(is it current?) 18:01
colomon pmichaud: nope
pmichaud maybe generate a list of rt tickets that have been closed since August? ;-)
colomon last updated in august
pmichaud that would certainly identify things fixed in nom that were broken in b
colomon errr, august 2010
pmichaud sorry, I meant NOMMAP 18:02
jnthn We really should update ROADMAP before the next relesae.
brb, phone
colomon ah, NOMMAP. (why isn't it in the docs directory?) 18:03
18:03 dakkar left
pmichaud it's intended to be temporary-ish, I believe 18:03
18:07 kaare_ joined 18:08 mj41 left, donri joined, Leo joined
Leo Who has the most advanced implementation of Perl6 at the moment? 18:09
PerlJam Leo: I do! 18:10
Leo: and you can too! :-)
Leo How do I get it then? 18:11
diakopter Leo: would you rather use mono/.Net or Parrot
Leo I'm on a mac, so that would be Parrot, right? 18:12
(Well, not on a mac right now... at work... but will be from home.)
18:12 hundskatt joined
diakopter mono runs on Mac 18:12
Leo Would rather not have to go the mono route. That the only way to get Perl6 on the mac? 18:14
18:14 sayu left 18:15 ksi joined
PerlJam Leo: no, the parrot version runs fine (or it did last time I tried it) 18:16
colomon parrot version ran fine yesterday. :) (haven't checked today's yet.) 18:17
(on mac, I mean)
18:18 birdwindupbird joined, birdwindupbird left
PerlJam colomon: btw, will your advent day 1 post really be about getting perl 6? so far it just describes the options. 18:19
Leo Can someone point me to a link or something for the latest parrot Perl6 then? I am sure I can Google it, but since I am here with all you knowledgable people.. maybe you guys know of a secret implementation or something. ;-)
diakopter depends how bleeding edge you want 18:20
PerlJam Leo: rakudo.org/how-to-get-rakudo-nom/
Leo PerlJam: Huh? This is a Perl6 forum right?
[Coke] if you want bleeding edge, and have git/perl/gcc installed, you can just grab the rakudo source from github and build from there, it will pull in everything you need.
PerlJam Leo: aye. Rakudo is the name of one of the Perl 6 implementations
japhb moritz, My apologies. :256[...] produces an Int. I forgot that you were trying to get a Buf. Although (big) Int -> Buf would be a nice operation to have.
Leo PerlJam: Thanks!
[Coke] ah, and that page does what I just said. PerlJam++ 18:21
18:21 am0c joined
pmichaud I think we no longer need the "git checkout nom" step 18:22
(doesn't hurt to have it, though)
18:22 am0c left
colomon PerlJam: yeah, once I decided to mention four implementations, explaining how to install each sounded like too much. But I should say something about the links, which do go to places with instructions. 18:22
pmichaud might also be worth changing "Building the nom compiler ..." to have "Perl 6" in it somewhere.
afk again 18:23
Leo More questions: How long (estimate) do you feel Perl6 (or Rakudo) will be "production ready"? Who (or what org) makes the official call when it's finally ready? 18:26
moritz Leo: the users need to make that call
PerlJam Leo: there is no "official" call on that and probably never will be.
moritz Leo: there are applications (critical avionic mission control) where the JVM isn't considered "production ready" 18:27
Leo What percentage of the Perl6 spec does Raukdo implement then? What's still missing? (Best answered with another link?) 18:28
PerlJam Leo: perl6.org/compilers/features.html :-)
Leo Moritz: Ha!
lichtkind will there this year perl6advent.wordpress.com/ ?
PerlJam lichtkind: yes! Want to write one of them? 18:29
lichtkind am not sure i know enough
yeah maybe
currently busy writing other stuff 18:30
i think 1 i can do
muixirt moritz: i like that notion of "production ready" :-) 18:31
PerlJam lichtkind: what's your email address? 18:32
lichtkind PerlJam: its my cpan ID 18:33
muixirt moritz: and what do they consider "production ready"?
dalek ecs: 3c53f87 | larry++ | S02-bits.pod:
Start moving Failure outside of Any
18:35 Vlavv` left
PerlJam lichtkind: you should get an invite soon (if not already) 18:35
18:35 Vlavv` joined
sorear good * #perl6 18:39
phenny sorear: 08:31Z <moritz> ask sorear if he'd like to write something for p6advent about his Unicode work for niecza
sorear slept through #phasers :| 18:40
colomon sorear: pretty quiet #phasers today anyway 18:41
sorear didn't we have a getting Perl 6 post last year? 18:45
snarkyboojum: still here?
colomon sorear: we had one two years ago, and the scene is quite a bit different these days. 18:47
mikemol This task looks like it might be interesting in Perl6: rosettacode.org/wiki/Paraffins 18:48
PerlJam seems to me it's been "getting rakudo" until now.
18:49 ksi left
PerlJam mikemol: write an advent post about it? ;) 18:49
mikemol If only I had time. My tuit count is at an all time low.
PerlJam seems to be the case for everyone. 18:50
mikemol nods
As soon as I have time available, I need to start jobhunting.
I have a job, but I need a different one.
Util mikemol: ooh, shiny! #Paraffins 18:52
18:53 molaf left, kaleem left 18:54 REPLeffect joined 18:55 MayDaniel joined, hundskatt left 19:01 ksi joined
colomon is doing his best to resist starting on the Paraffins thing.... 19:05
TimToady I'd've called it "alkanes" myself... 19:07
especially if we eventually get "alkenes" and "alkynes" 19:08
sorear it's an abstraction anyway, it doesn't handle steric hindrance 19:14
the name has historical significance iirc 19:15
lichtkind PerlJam: got it 19:22
TimToady is inclined to think that Failure objects no longer need a .handled state 19:24
benabik If Failure is outside of any would sub epic($x) { $x }; epic fail "test" actually epically fail? 19:25
Oops.
Wait, no that was "right"
moritz epic() would never be called
TimToady no, fail returns
benabik Awh. 19:26
jnthn It'd pass CHECK though :)
moritz because the fail() returns the routine that the code is in
TimToady is why I suggested epic-fail
benabik Well, I guess it would be an epic failure, just not the kind I was expecting. :-D
I guess epic would have to be some kind of macro. :-D 19:27
TimToady maybe a statement_prefix 19:28
lichtkind PerlJam: now im in 19:32
PerlJam lichtkind: now all you need to do is claim a day and write something :) 19:34
19:36 MayDaniel left
[Coke] wonders what the point of skipping a test that is passing for the wrong reasons is. 19:36
Surely it's better to let it run and at least then you'll find out if it starts failing. 19:37
moritz I usually skip them
[Coke] #?rakudo skip 'RT81718 (false positive in nom)'
moritz: how do you know when to unskip them? 19:38
you can't just autounfudge them, obviously.
moritz [Coke]: by running it occasionally, and checking if the error is right
I know that's LTA
19:38 wormphlegm joined
[Coke] moritz: er, what error? 19:39
moritz isn't that about an error?
[Coke] If it's /passing/ ... what error are you checking?
lichtkind PerlJam: still wresle with the interface
moritz isn't that a dies_ok?
[Coke] yes.
it's dying. that's what the test is checking. If that's not sufficient, shouldn't we rewrite the test? 19:40
moritz yes, we should
[Coke] and then TODO it?
ok. Thanks. ;)
moritz but checking the error message is also fragile
which is why I'm working on my exceptions grant :/
[Coke] gives up on autounfudge. too much work. 19:41
dalek p: e680f11 | moritz++ | tools/build/PARROT_REVISION:
bump parrot revision
19:42
kudo/nom: 2b31355 | moritz++ | / (2 files):
implement input-line-separator on IO::Socket::INET (not yet tested, but not regressing either)
19:43
19:46 hundskatt joined
[Coke] Do we have anyway to fudge based on anything other than platform? There's at least one thing that is fudge-skipped with a "fails on windows". 19:52
TimToady that seems like a rathole 19:53
moritz [Coke]: I'd rather not do that
[Coke] Am i the only one annoyed by skipping passing tests? ;) 19:54
especially with all the regressions we just did on purpose, I'd hate for more to slip through accidentally. 19:55
moritz I'm much more annoyed by passing TODOs that clutter the test result report
[Coke] moritz: yes. I'm suggesting we avoid BOTH of those, and only todo the test when it is known to fail, not skip it when it might fail.
19:56 daxim left
benabik [Coke]: Are you talking about S16/unlink? I notice that there's a $iswin32 variable that goes completely unused. 19:57
moritz [Coke]: it's not (yet) a pain point for me
dalek ast: aeb8722 | coke++ | S02-literals/numeric.t:
Unfudge passing test (RT #70600)
19:59
[Coke] nom: my $a="100.B".."102.B";say $a;say "$a"; 20:05
p6eval nom f0ac58: OUTPUT«"100.B".."102.B"␤No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu, Mu %_)␤␤ in method Numeric at src/gen/CORE.setting:634␤ in sub infix:<==> at src/gen/CORE.setting:2099␤ in sub infix:<==> at src/gen/CORE.setting:2099␤ in…
[Coke] nom: my $a= +"100.B"; 20:06
p6eval nom f0ac58: ( no output )
[Coke] nom: my $a= +"100.B"; say $a; say "$a"
p6eval nom f0ac58: OUTPUT«Failure.new()␤Decimal point must be followed by digit␤ in method Numeric at src/gen/CORE.setting:3131␤ in sub prefix:<+> at src/gen/CORE.setting:1911␤ in block <anon> at /tmp/mcndu7TeBT:1␤ in <anon> at /tmp/mcndu7TeBT:1␤»
[Coke] do we have a bot announcing new rakudobugs? 20:11
20:12 Chillance left, wormphlegm left
PerlJam
.oO( How would the bot know about new bugs? )
20:15
;-)
jnthn Just use masak as a bot. :) 20:16
20:16 abercrombie left 20:17 abercrombie joined
moritz nom: say 0e9999999999999 20:18
p6eval nom f0ac58: OUTPUT«NaN␤»
[Coke] nom: say 0e9999999999999 == 0 20:19
p6eval nom f0ac58: OUTPUT«Bool::False␤»
[Coke] Odd. works for me.
wonder if it needs a bignum library that the bot doesn't get.
PerlJam nom: say 1e9999999999999;
p6eval nom f0ac58: OUTPUT«Inf␤»
benabik Maybe the NeM notation isn't using bignums? 20:20
PerlJam nom: say 0e999; say 1e999;
p6eval nom f0ac58: OUTPUT«NaN␤Inf␤»
benabik nom: say 1*10^9999999999999
moritz benabik: it isn't for the exponent, I think
p6eval nom f0ac58: OUTPUT«one(10, 9999999999999)␤»
moritz nom: say 0 * 10 ** 9999999999999
benabik nom: say 1*(10**9999999999999)
p6eval nom f0ac58: OUTPUT«NaN␤»
nom f0ac58: OUTPUT«Inf␤»
benabik Strange.
moritz not so strange
[Coke] swears this was working locally. 20:21
moritz benabik: just think about how much storage 10 ** 9999999999999 would need
benabik … Details
I want to shoot myself in the foot, darn it. ;-)
PerlJam perl should stringify that number to "Ridiculously large" :) 20:22
colomon it shouldn't need more than "10 ** 9999999999999".chars bytes, right? ;)
moritz I've intentionally mapped bigint ** bigint to Inf
colomon: :-)
[Coke] nom: say so 0e9999999999999 == 0 20:23
p6eval nom f0ac58: OUTPUT«Bool::False␤»
moritz though maybe we should special-case the 0e99999999999 thing
[Coke] nom: say so 0e999999999999999 == 0
p6eval nom f0ac58: OUTPUT«Bool::False␤»
[Coke] that line, which is the one from the ticket, DOES work locally.
benabik 0e<don't care> = 0 ?
moritz yes 20:24
[Coke]: woah
PerlJam sounds sensible to me.
20:25 am0c joined
benabik nom: (2**65).WHAT.say 20:25
p6eval nom f0ac58: OUTPUT«Int()␤»
20:25 am0c left
PerlJam
.oO( We've secretly replaced Coke's version of Rakudo with one from the future. Let's see if anyone notices )
20:26
[Coke] Regarding 105004; "nothing else is derived" != "Nothing else CAN be derived from it".
benabik nom: my $x = 2**65**0; $x.say; $x.WHAT.say
p6eval nom f0ac58: OUTPUT«2␤Int()␤»
benabik nom: my $x = (2**65)**0; $x.say; $x.WHAT.say
p6eval nom f0ac58: OUTPUT«1␤Int()␤»
20:28 overrosy left 20:30 overrosy joined 20:36 localhost left 20:37 localhost joined
[Coke] TimToady: should users be able to subclass Whatever if they want? 20:40
masak colomon: both mberends and [Coke] have also expressed interest in hacking on Pugs. 20:41
jnthn nom: class ScrewItIDon'tCare is Whatever { }
p6eval nom f0ac58: ( no output )
colomon masak: it is fair to describe you and ingy as ringleaders, though, right?
[Coke] masak, colomon: it is SUPER unlikely I will find time for that. 20:42
jnthn [Coke]: Is there a bug in doing so that makes you ask? :)
[Coke] jnthn: there is a bug just opened that you CAN subclass it.
jnthn [Coke]: Huh. 20:43
Why shouldn't you be able to...
:)
[Coke] and the submitter claims this is a bug because the spec says "nothing derives from it"
masak colomon: sounds like it's a sect, not a restoration project. but beyond that, I guess it's fair :)
[Coke] so I'm just hoping we can clarify the spec and close the rakudobug.
jnthn [Coke]: Do they give an actual spec ref?
[Coke] rt.perl.org/rt3/Ticket/Display.html?id=105004 20:47
cognominal_ is ingy a european to be diacritical?
[Coke] which references S04:1292
I think he's just a contrarian.
masak cognominal_: he's a döt on the 'Net, but not a cog in the machine. 20:48
[Coke] (ingy, not stephane) 20:49
cognominal_ I am a cog too :)
masak .oO( cogtrarian )
20:50 carlin joined
dalek ast: b75d873 | coke++ | S03-operators/range.t:
fudge RT#105014 for rakudo
20:51
cognominal_ I am using pir:: and get burnt using such an instruction as last statement of a function. I think it is not the first time. jnthn, should I file a bug?
I am not sure if these internals should be documented or supported. 20:52
[Coke] don't use pir::
are you using it in rakudo code or nqp code?
benabik [Coke]: Doctor, it hurts when I use pir::?
[Coke] if rakudo, I'd say don't do that. 20:53
... I'm biased, though, as I'm trying to keep the # of tickets down. ;)
jnthn cognominal_: Typical users should not be using pir::. The only reason you don't need a pragma to be allowed to use it is because I didn't get around to it. 20:54
cognominal_ I am dabbling at undestanding what is necessary for a f:rom<nqp> so I am not sure how to answering coke.
jnthn cognominal_: If you are using it, then there's a various things you need to know, and many, many ways to make the code-gen go wrong.
cognominal_ jnthn, I have already hit quite a few of them :) 20:55
jnthn cognominal_: One of them is that if you use a void op as the last thing in a block, you're going to have fun :)
Anyway, I don't think we should have tickets about pir::. We could end up with a dozen tickets about it. 20:56
cognominal_ that was implicitely my point when I said [21:51] <cognominal_> I am not sure if these internals should be documented or supported.
on the other hand I find pir:: code a good way to have a feel about the internals. 20:57
jnthn cognominal_: Documented is fine if it's for the benefit of Rakudo developers.
cognominal_: Though really we should be using nqp:: ops these days.
cognominal_: But I'd strongly discourage their use out of core. 20:58
cognominal_ I agree.
20:58 araujo left, overrosy left 21:00 overrosy joined
dalek ast: 1c4c5f0 | coke++ | S06-other/main-usage.t:
fudge RT#105018 for rakudo
21:01
21:02 ksi left, ksi joined, araujo joined, araujo left, araujo joined
[Coke] jnthn: FYI, I'm going to start giving tickets to things that are marked "nom regression" 21:07
jnthn [Coke]: ok, wfm.
[Coke] (any new ones I'm finding in "make spectest" are getting tickets outright.)
lichtkind PerlJam: i still dont grok the system but i think writing something would be good anyhow :) 21:09
21:10 bluescreen10 left, y3llow left, y3llow joined 21:18 overrosy left 21:20 overrosy joined 21:39 dorlamm joined 21:53 wolfman2000 joined, alester joined, prammer left 21:58 ethndbst left
masak [Coke]++ # RT-ifying nom regressions 21:59
22:18 kaare_ left 22:19 dorlamm left 22:33 ksi left
sorear good * #perl6 22:37
22:40 prammer joined 22:41 whiteknight joined, whiteknight is now known as Guest75643
masak good night, #perl6 22:42
sorear bye 22:43
colomon o/
22:43 overrosy left 22:45 overrosy joined 22:46 lichtkind left 22:52 overrosy left 22:54 overrosy joined 23:04 benabik left, muixirt left 23:15 overrosy left 23:16 vmspb joined 23:17 overrosy joined 23:18 awwaiid left 23:24 overrosy left, buubot_backup left 23:25 buubot_backup joined, huf left 23:26 overrosy joined 23:27 huf joined 23:32 awwaiid joined 23:34 envi_ left, cooper joined 23:35 wolfman2000 left
vmspb Hi, sorear. Niecza does not have built-in Readline? 23:51
I used on Ubuntu an external Readline shell - RLFE. It looks good
$ rlfe mono run/Niecza.exe or
$ rlfe -h ~/.perl6_history mono run/Niecza.exe
sorear vmspb: I use rlwrap with the niecza repl personally 23:58