»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend! | feather will shut down permanently on 2015-03-31
Set by jnthn on 28 February 2015.
Mouq raydiak: We already removed a lot that were parrot-only, but there are still more 00:04
raydiak: Re: prioritization, there is a section in "edit search" to change that… and I think it's a very good idea to look into applying it 00:06
ShimmerFairy I still wonder how helpful it may be to have our own separate RT instance for rakudo. (At least, it feels interesting to me to have P5 and P6 bugs share the same space)
Mouq ShimmerFairy: Well it might be helpful with organization. There's a lot of different regions of the Perl 6 language and the Perl 6 compiler, and we really don't (can't?) organize our tickets into these regions effectively 00:08
ShimmerFairy I know we've talked of moving before, and I thought "our own RT" might be a happy medium between "keep RT" and "use github issues" 00:09
(and maybe we could use something other than bitcard, something that actually supports Unicode :P)
timotimo raydiak: allegedly angular 1.3 got a whole lot of performance improvements over 1.2; i'm not sure which version we're using for the profiler 00:10
raydiak would strongly prefer to be using github issues, as he has several problems with rt usage including repeated mysterious breakage of his account where he does not have access to view bugs when he is logged in, which was supposedly fixed by [Coke]++ once or twice alraedy
00:10 retupmoca joined
timotimo but maybe updating that can lead to a nice performance boost? 00:10
raydiak timotimo: yep good call, looks like 1.2.21
ShimmerFairy raydiak: perhaps that could be fixed by not using bitcard? (AFAIK bitcard is not a part of RT)
Mouq timotimo: Err, that's about Angular, not RT, right? 00:11
raydiak ShimmerFairy: I didn't see another option when I went to create an account
Mouq oh lol
timotimo yeah, we're using angular 1.2 there
Mouq sorry, I lost the thread of the convo :)
ShimmerFairy raydiak: yeah, but if we were to set up our own RT, we could choose to not use bitcard to handle accounts for the system :)
timotimo i have no idea how uch would have to change to get from 1.2 to 1.3
all i know is for big-ish profiles, i have to click "don't kill this tab yet, please" twenty times and then chrome says "oops, this tab crashed!" anyway ;( 00:12
ShimmerFairy I used to be against RT too, but not so much anymore. I appreciate strongly that it lets you submit bugs without an account, and if we would just add in a web-based "new bug" entry form (instead of only having email), it'd be fine by me.
00:13 vendethiel left
raydiak timotimo: yeah half or more of what I have on the ecosystem brings it to its knees 00:14
even in the smallest test runs
(considering that raytracing a single pixel doesn't give you much of a functional test) 00:15
timotimo not that terrible for me, but i often do things with not too many different routines
Mouq ShimmerFairy: Also, I have no idea how to/if you can do formatting on RT 00:16
ShimmerFairy I will say though, that github issues doesn't look/feel as powerful as RT or other systems. I could easily be wrong on that, though. 00:17
raydiak timotimo: I've been trying to do things in nice manageable chunks, but apparently most of what I want to do just necessarily involves nested potentially long-running loops and so forth
Mouq ShimmerFairy: OTOH, think there's a lot to be said for the power of good UI :P
raydiak not that much of that couldn't be at least improved by a marginal reduction in my laziness :)
ShimmerFairy Perhaps instead of angularjs, you could turn to server-side scripting, like *cough* Perl 6? :) 00:19
Mouq: it's certainly not a binary choice between RT and GI, but I do think we need our own, separate bug tracker. (And I also still feel like GI doesn't have as much manipulexity as other systems, and perhaps not enough for our needs) 00:20
00:24 laouji joined
Mouq ShimmerFairy: +1 00:25
00:26 tinyblak joined
ShimmerFairy I feel like something that lets you submit bugs without an account would be nice (esp. considering we're working on a compiler, so we could reasonably assume lots of people will only ever have one bug to report), but I don't know how many bug trackers do that. 00:28
I also don't know if other people here like the sound of that so much :)
00:38 vendethiel joined 00:39 Sir_Ragnarok joined, laouji left, Sir_Ragnarok_ left 00:40 laouji joined
raydiak afaik we have no desire to limit or prevent bug reports, though I suspect such a thing could be accomplished on github too (albeit with some extra work on our own e-mail relay and/or submission form) 00:43
00:44 laouji left 00:45 laouji joined
raydiak wrt js (angular or otherwise), I am not so sure about requiring a local webserver just to view the results of your profile run, but it is possible that we might be able to move more of the processing from the js to the nqp which builds it 00:47
dalek ast: 0cc81eb | Mouq++ | S32-io/io-handle.t:
Add test for RT #123888
 00:48
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=123888
raydiak I'll know more when I look in detail; today is mostly just having a strategy meeting with myself
Mouq It's unfortunate that github.com/camelia is taken :9 00:49
raydiak heh you saw right where I was going with that :) kinda wonder if it isn't one of us who reserved it 00:52
timotimo the biggest problem with our profiler data blobs is currently that we put the complete call tree in there 00:54
it's potentially ginormous 00:55
there's a branch in which i tried to "compress" the call-tree as it's being filled with data by the low-level moarvm-internal profiling instructions
but it gives wrong results and i haven't had the tenacity to figure out what wrongs it
raydiak compress as in get rid of unneeded data, or as in lossless compression? 00:57
00:57 yqt left
timotimo collapsing all nodes below a given node into a single node 00:58
before creating the nodes below that note
raydiak oh yes I remember this
timotimo "finite_callgraph_depth" is the name of the branch
oh, my explanation was wrong 01:00
this happens after the nodes have been created
01:00 breinbaas left
timotimo the code runs where the internal data is turned into the data structure that is going to be json-dumped later on 01:00
raydiak it'd definitely be nice to have, though I don't think there is going to be any getting around the need to make the JS more robust in the long run 01:02
but...gotta run, sorry to ditch out in the middle of this :) 01:03
o/ #perl6
retupmoca .botsnack
yoleaux :D
timotimo it doesn't seem like any breaking changes between 1.2 and 1.3 would have an impact on our profiler app 01:04
bye raydiak :)
hello retupmoca
retupmoca hello 01:06
timotimo it seems like just changing from 1.2 to 1.3 makes sorting by exclusive time more than twice as fast 01:11
however, the pages of the profiler are no longer exclusively visible, they are now all visible at the same time
likely due to a change in expressions?
ng-show="Tab == 'Overview'" is how we did that 01:12
dalek ast: 1aa7d9d | Mouq++ | S05-grammar/parse_and_parsefile.t:
Add a test for RT111768
timotimo errors.angularjs.org is down, and so i cannot figure out what this exception is supposed to mean >:( 01:15
01:17 virtualsue left 01:26 vendethiel left 01:36 laouji left
TimToady Mouq: that test is...deliciously perverse :) 01:39
01:44 pecastro left 01:45 vendethiel joined 01:47 laouji joined 01:50 redhands joined 02:02 virtualsue joined 02:06 tinyblak left, vendethiel left, tinyblak joined 02:11 tinyblak left 02:12 vendethiel joined
japhb .ask jnthn In your OO::Monitors module, is there any attempt to support inheritance heirarchies, in which all classes are monitors? In other words, can it handle: monitor A { method foo }; monitor B is A { method bar }; and keep foo and bar mutexed? 02:27
yoleaux 7 Mar 2015 10:39Z <jnthn> japhb: $*SCHEDULER.uncaught_handler = -> $ex { #`( do something ) }
japhb: I'll pass your message to jnthn.
japhb .tell jnthn Yeah, shortly after I .asked you, someone (timotimo?) pointed out the workaround of setting uncaught_handler *after* creating the new $*SCHEDULER, because of the "BUILD from a composed Role gets smashed" problem. 02:30
yoleaux japhb: I'll pass your message to jnthn.
japhb .tell jnthn But even so, I'm unable to figure out how to *trigger* an uncaught exception that .uncaught_handler *does* catch. 02:31
yoleaux japhb: I'll pass your message to jnthn.
japhb Man, I'm really tired of TZ skew.
02:31 colomon left 02:32 colomon joined, virtualsue left 02:34 vendethiel left 02:39 kaare__ joined 02:43 kaare_ left 02:52 tinyblak joined 02:59 kaare__ is now known as kaare_ 03:24 vendethiel joined 03:26 noganex_ joined 03:29 noganex left 03:39 colomon left, colomon joined 03:50 Ben_Goldberg left
ugexe then ill just fatpack that bitch into a standalone 04:03
oops...
04:03 Sqirrel left 04:07 vendethiel left
avuserow .ask FROGGS I'm looking at a bug in nqp-j (github.com/perl6/nqp/issues/224) and saw you did some work with eof. I'm testing a patch, but was curious if there's any reason you didn't use fc.size() rather than getc? 04:12
yoleaux avuserow: I'll pass your message to FROGGS.
04:22 vendethiel joined 04:30 redhands left 04:41 molaf left 04:47 vendethiel left 04:55 raiph left
raydiak drops a pin 05:03
TimToady ow! 05:04
raydiak picks up said pin and offers TimToady a band-aid with apologies for his carelessness
avuserow o/ raydiak 05:06
raydiak heya avuserow 05:07
avuserow how's it going?
avuserow is waiting on a jvm spectest :|
raydiak eh, won't complain, everything I'm working on atm is going fairly well...how about you? 05:08
TimToady is still failing lots of JVM tests
raydiak due to...???... 05:09
avuserow just hopping around on various ideas, trying to see where I get stuck and reporting bugs as possible
yeah, JVM looks pretty unhappy so far for me :(
raydiak has been worrying about the state of our jvm backend vs tuits
TimToady only fails about 64 test files... 05:10
well, some of them are unexpectedly succeeding TODOs 05:11
avuserow both jvm and moar have distinct failure modes for my current idea. the JVM issue appears fixable with my skillset though
raydiak avuserow: what is your current idea? 05:12
avuserow I have a lot of large .sql backups from an old project, and I want to convert it to an incremental format to deduplicate some data. First step is parsing 4-5 GB .sql files.
(this being the more fun course of action, rather than just adjusting my data hoarding habits, of course) 05:13
raydiak ow...I can feel your pain from here...large datasets seems to be one of our weaknesses for the time being 05:14
05:14 vendethiel joined
avuserow I think the JVM can parse about 1 file in a half hour or so, which is viable 05:15
moar would be faster but I run out of memory
raydiak (faster :) )
avuserow and once I get things working, I can potentially improve the grammar 05:16
raydiak how so?
avuserow one thing I have to match is a string that may include backslashed quotemarks. I had a really difficult time figuring out how to do this, so my strategy might be wasteful
raydiak may I see? 05:17
avuserow gist.github.com/avuserow/3e02faf868f743de03ef 05:18
TimToady [ <-["]> | \\'"' ]* is the normalish way
though might be faster with || 05:19
avuserow I'm hoping to make a full-ish SQL grammar someday
TimToady since most chars are not \\"
ShimmerFairy TimToady: I assume it's <-["]> and not . to account for a closing " ?
TimToady er, <-[\\"]> maybe 05:20
if there's anything else that can be backslashed, the <-\\"]> | \\(.) or so
avuserow somehow I was able to have a simple-string rule and a string-with-escapes rule, and it picks the right one... not sure if that works universally, but it made the smaller benchmarks about 3x faster to have both rules 05:21
TimToady er, only spelled right
or you can do [ <-[\\"]>+ || \\(.) ]* with the + to collect more non-specials in one match 05:22
avuserow I tried something like that but couldn't get it to work. I'll have to try that again 05:23
ShimmerFairy I would probably do / ['\' <[nt\\'"]> | .]* / or similar (for backslash sequences of \n, \t, \\, \', and \")
avuserow but yeah, actually writing the grammar was a fun experience. obviously I didn't start with a formal definition of SQL, which I plan on revisiting, but it was still nice how fast I got the basic parser up and running 05:24
once I get a full end to end parse on this file, I'm planning a writeup 05:25
TimToady doing it with . means you're definitely relying on LTM, so || wouldn't work 05:27
well, it might in that order, but certainly not in the other order
but if someone backwhacks something you're not anticipating, the . could end up eating a single backslash, which might be surprising
raydiak worked on something long long ago in his abused intern years, never released, called morphbase, which grappled with similar issues and several more...though that was written in core p5, no grammars/Parse::RecDescent/etc 05:29
but in any case, it relied on even vs odd number of backslashes
b/c the even cuonts can always be ignored as N number of literal Q«\» 05:30
*counts
avuserow I wrote a p5 script to change strings down to a single letter, and had several issues with getting _that_ right
ShimmerFairy TimToady: sure, it's hard to write something like that in a general sense, without letting some potential issues through. 05:31
05:32 Sqirrel joined 05:43 CPiral joined 06:00 tinyblak left 06:05 Ugator joined
avuserow \o/ parsed the file completely after patching a bug in nqp-j. now to submit a PR, and refactor to actually use the information 06:14
06:15 vendethiel left 06:21 mr-foobar left
masak good morning, #perl6 06:28
06:32 MadcapJake left, tinyblak joined 06:38 rindolf joined 06:39 dayangkun joined 06:47 perl6_newbee joined
perl6_newbee morning or night respectively 06:47
06:50 dayangkun left
[Tux] Mouq, awake? 06:57
06:59 vendethiel joined
perl6_newbee all are sleeping :-) 07:00
[Tux] .tell Mouq gist.github.com/Tux/3efcf2bbcb859a27ef6a <= RT#123888 is incomplete. reopen? 07:01
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=123888
yoleaux [Tux]: I'll pass your message to Mouq.
[Tux] note that there is no newline *at all* in my example, incontrast to the example that shows the current fix 07:02
07:04 prime left 07:05 rurban joined 07:06 prime joined
perl6_newbee [Tux]: when I want to make a sugegstion, can I create a ticket via [email@hidden.address] with the tag [SUGGESTION] or should I use [bug]? What do you think? 07:09
[Tux] If I am unsure (like you are) I ask here, and I am a perl6 newbee too
perl6_newbee :-) 07:10
[Tux] I am a seasoned perl5 programmer
feeling like a bulldozer trying to map my thoughts into this new language
perl6_newbee Hmm perl5 is cool and I think it should not be too hard to understand perl6 with perl5 skills, is it? 07:11
[Tux] right *and* wrong at the same time
it is a completely different language
perl6_newbee agree. I have been playing with perl6 for about one month and I like perl6 even more than perl5 because it does many things better 07:13
07:14 colomon left 07:15 colomon joined 07:19 nwc10 joined 07:21 Sqirrel left
masak perl6_newbee: maybe start by making the suggestion here? perl6-language is another possible forum. 07:22
(I usually start here, then go to p6l if the issue seems "big" enough.)
perl6_newbee hi masak. 07:23
masak hi hi
07:23 vendethiel left
perl6_newbee The topic is the BOM issue I raised last friday here. I read the commecnt from jnthn and got the feeling it would be possible. so I thought I have to create a ticket in order to prevent it from forgetting. 07:24
masak don't know exactly what that's about 07:25
but if it's something that could be improved, then it probably deserves to be in RT
perl6_newbee I encountered the problem on Windows that the Notepad.exe saves the BOM at the start of the text. Perl6 read in the file but does not remove the BOM flag 07:26
That is not intuitive from my point of view. If I have to handle this flag for myself
07:27 wicope joined
perl6_newbee so I ask here and at least jnthn said it could be implemented at the right place without side effects. 07:27
I created a ticket. If it is not important or wrong someone will close it, I think. 07:28
07:34 telex left 07:36 telex joined
masak it removes it when it's the source code, I'm pretty sure. 07:38
but a good argument could be made that it should remove it for (text) input files, too.
perl6_newbee++
07:40 tinyblak left
Mouq [Tux]++ 07:41
yoleaux 07:01Z <[Tux]> Mouq: gist.github.com/Tux/3efcf2bbcb859a27ef6a <= RT#123888 is incomplete. reopen?
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=123888
Mouq I'll reopen and update the ticket :)
[Tux] thanks
tux.nl/Files/20150309084206.png <= this is what made me re-check 07:42
perl6_newbee masak: :-) 07:43
Mouq [Tux]: It doesn't seem to be about the newlines, it seems to be about supplying perl6 with a file for STDIN instead of directly to STDIN 07:47
[Tux] ok
so my bug report was actually touching TWO buglets? 07:48
DrForr For those who may be concerned: rakudo.org/how-to-get-rakudo responds with a 500 error over Firefox.
07:49 tinyblak joined 07:51 ggoebel left, tinyblak_ joined, ggoebel joined 07:52 tinybla__ joined
Mouq DrForr: And so everyone on the channel visits that page :P 07:52
(DrForr++ of course :) 07:53
DrForr Well, of *course*. It's the Internet equivalent of a 'WET PAINT' sign. 07:54
07:54 tinyblak left 07:56 tinyblak_ left, Ven joined
Mouq And [Tux], I don't think so… I'm pretty sure it was one bug disguised as another, in this case something going wrong with $*IN when it's secretely a filehandle 07:57
07:57 zakharyas joined
[Tux] whatever, thanks for looking into them and fixing 07:57
07:58 [Sno] left
Mouq For now, sleep & 07:58
[Tux] thinks that in 5 month /me has TWO favorite languages
if only speed speed speed
DrForr Ah, it's accessible again. 08:00
dalek p: a27f52e | (Adrian Kreher)++ | src/vm/jvm/runtime/org/perl6/nqp/io/FileHandle.java:
Fix EOF with newlines on 32k byte boundaries

Fixes #224 by changing the EOF checking to use the size of the file rather than attempting to read a character using getc(). This fixes a problem which happens when a line ends on a multiple of 32768 bytes.
p: a0dbe4f | FROGGS++ | src/vm/jvm/runtime/org/perl6/nqp/io/FileHandle.java:
Merge pull request #225 from avuserow/master

Fix EOF with newlines on 32k byte boundaries
08:02 FROGGS joined 08:04 darutoko joined 08:08 larion left 08:09 vendethiel joined 08:14 Rounin joined 08:18 Ven left 08:19 dayangkun joined, dayangkun left 08:22 Foxcool joined
perl6_newbee cu guys, gotta go 08:24
08:25 perl6_newbee left 08:33 vendethiel left 08:35 Ven joined, Ven left, CPiral left
DrForr 'my $i = UndeclaredClass.new;' properly throws an error, but it includes 'UndeclaredClass used at line 3' instead of 'UndeclaredClass used\nat /full/path/to/file.t:3'. Another bug? 08:36
08:37 virtualsue joined 08:38 kaleem joined 08:39 jdv79 left
DrForr (moarVM on latest rakudo) 08:41
08:44 Ven joined, abraxxa joined
FROGGS m: my $i = UndeclaredClass.new; 08:45
yoleaux 04:12Z <avuserow> FROGGS: I'm looking at a bug in nqp-j (github.com/perl6/nqp/issues/224) and saw you did some work with eof. I'm testing a patch, but was curious if there's any reason you didn't use fc.size() rather than getc?
camelia rakudo-moar cd7c31: OUTPUT«5===SORRY!5=== Error while compiling /tmp/GeW8HECuxX␤Undeclared name:␤ UndeclaredClass used at line 1␤␤»
FROGGS DrForr: yes, it is a bug that -e in camelia here is not mentioned
08:46 vendethiel joined
DrForr Should I rakudobug that as well? 08:46
08:51 robinsmidsrod left 08:53 [Sno] joined, robinsmidsrod joined
FROGGS DrForr: yes, please do 08:54
08:56 wicope left
torbjorn is there any chance we could have the perl6 eval bot in #oslo.pm at irc.perl.org? 08:57
FROGGS torbjorn: I already talked to moritz++ about that
but, let me try something...
torbjorn oh cool
moritz you don't already?
DrForr Will do. 08:58
torbjorn same nick on irc.perl.org?
moritz torbjorn: p6eval on magnet, iirc
08:58 wicope joined
dalek albot: 0756169 | FROGGS++ | irc.perl.org.conf:
send camelia to #oslo.pm
 08:58
FROGGS p6eval: restart 08:59
moritz FROGGS: won't work that way
FROGGS: have to pull first :-)
torbjorn its in oslo, not oslo.pm
FROGGS evalbot: control pull 09:00
torbjorn oh its in the git pull
FROGGS evalbot: control restart
hmmmm
moritz FROGGS: don't
FROGGS: that's the wrong instance you're trying to restart
moritz does it on the command line
FROGGS ohh, I thought there is just one :o)
moritz nope, separate process per server 09:01
FROGGS moritz++
Ven o/, #perl6
moritz Mar 09 10:01:20 ns1 bash[1059]: Can't read configuration in perl.org.conf: No such file or directory
FROGGS hi Ven
moritz dumb
09:02 MilkmanD1n joined
FROGGS it does not like the dot? 09:02
moritz no
FROGGS ohh, it is called irc.perl.org.conf
moritz I tried to start the wrong instance
09:03 dustinm`_ joined 09:04 MilkmanDan left, dustinm` left 09:09 vendethiel left 09:10 Celelibi left, MAB90RX joined 09:11 djanatyn left, kjs_ joined, djanatyn joined 09:13 isBEKaml joined 09:14 larion joined 09:15 fhelmberger joined
masak (camelia going places)++ 09:16
09:18 isBEKaml left 09:21 vendethiel joined
DrForr Error report sent, perl #124027 filed. 09:24
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124027
DrForr Again pretty trivial, just basically making stupid mistakes and seeing if the errors make sense :) 09:26
09:28 Ugator left, dakkar joined 09:30 tinybla__ left 09:33 virtualsue left 09:36 larion left 09:39 Celelibi joined
Ulti can I suggest comments are removed and turned off for the majority of rakudo.org pages? 09:48
stuff like the bottom of this page for example just look really bad rakudo.org/how-to-help/
moritz +1
09:50 smls joined 10:03 espadrine_ joined 10:04 tinyblak joined 10:05 vendethiel left 10:10 rurban left 10:17 virtualsue joined 10:18 laouji left 10:19 coffee` joined 10:22 cschwenz joined, vendethiel joined 10:23 cschwenz left 10:32 donaldh joined
donaldh lizmat++ for looking at JVM but it's still broken... 10:33
yoleaux 7 Mar 2015 11:31Z <lizmat> donaldh: seems like I fixed the JVM build
10:34 rurban joined
FROGGS donaldh: what is broken? 10:36
donaldh I'm still getting lots of test failures, as also reported by TimToady++ 10:38
FROGGS donaldh: spectest failures or module test failures? 10:39
donaldh And it looks to me like the culprit is the nqp::index fix I applied, to unbust TimToady's grammar change on JVM.
spectest failures
FROGGS ohh 10:40
then we should fix the fix :o)
10:40 kaleem left
donaldh Well I can't just back out the fix in nqp/jvm, without also applying a workaround in the rakudo grammar 10:41
First I'll spend some tuits to see if I can understand the regression in nqp/jvm
It looks like something in nqp's JVM backend was relying on the broken behaviour of nqp::index 10:42
FROGGS hmmm, that sounds hard to believe
moritz and the broken behavior was to throw an exception? 10:44
or return the wrong value?
ab5tract lizmat++ # i was actually working an an s/at_pos/at-pos/ patch on friday night, but i like the the all-caps variant even better! 10:45
10:45 vendethiel left 10:46 Woodi left, kaleem joined, Woodi joined
donaldh I think it was throwing an exception that was being caught in nqp. Seems to cause behaviour change in the resulting runtime, regardless of the nqp::index behaviour in that runtime. 10:46
10:48 larion joined 10:52 vendethiel joined
|Tux| WTF. Who is to blame for test-x 50000 35.280 35.059 10:55
yoleaux 7 Mar 2015 18:33Z <nine_> |Tux|: the performance regression is not caused by Inline::Perl5 but by Rakudo. I suspect it's jnthn++'s native work.
|Tux| not only first time below 36, but way below 36 10:56
you guys rock!
masak we're only getting started!
|Tux| I still need a factor 900 to match XS though
904.615 10:57
masak a noble goal.
|Tux| :)
moritz |Tux|: do you have a comparison to a pure-perl implementation? 10:58
|Tux| gist.github.com/Tux/82b5aa66cc168c90ebd3 10:59
top 6 are perl5
rest is perl6
ask!
test-pp is Text::CSV_PP which is the pure-perl5 version of Text::CSV_XS 11:00
moritz why is csv-easy-pp so much faster than csv-test-pp?
|Tux| the first two are included for compare only: they do not support options and crash on every invilid CSV
test-pp has all the options that test-xs also supports. Checking options costs performance 11:01
moritz m: say 38.777 / 6.793
camelia rakudo-moar cd7c31: OUTPUT«5.708376␤»
moritz m: say 38.777 / 0.513 # that's the one that actually interests me, I guess 11:02
camelia rakudo-moar cd7c31: OUTPUT«75.588694␤»
moritz pure p5 vs. pure p6
|Tux| that one is a state machine in perl6 that has NO options and also does not support multi-character EOL or SEP
the 6.7 version that is
11:03 yqt joined
moritz but apples:apples comparison is 38.777 vs. 0.513, right? 11:03
|Tux| so yes, pure-perl vs pure perl6 is a factor 75
moritz not good, but we've done much worse on that in the past
|Tux| yes, first measurements were about 260 seconds :)
so having 35 now is GOOD 11:04
don't compare to test. compare to test-x
moritz is test-x merged in master?
moritz doesn't even know what test-x is, just assumes it's a branch 11:05
|Tux| test is a reference for myself: it is to ensure that test-x does not slow down compared to test
test-x is my working version of Text::CSV on github.com/Tux/CSV
11:07 virtualsue left
|Tux| FWIW the ip5 versions will probably go back to around 8.2 again (Inline::Perl5 versions of test-xs and test-pp) 11:09
11:15 vendethiel left 11:19 vendethiel joined 11:21 virtualsue joined 11:23 coffee` left
dalek p: 1dfa4ca | donaldh++ | src/vm/jvm/QAST/Compiler.nqp:
Unbust nqp on JVM

Was relying on nqp::index returning 0 for empty search string.		 11:24
11:29 diana_olhovik_ joined 11:35 yqt left
colomon .u ° 11:41
yoleaux U+00B0 DEGREE SIGN [So] (°)
donaldh CallFrame on JVM sometimes returns with no 'line' annotation. It looks like Moar always returns 'line' => "1" in these cases. 11:43
This makes Test.pm noisy on JVM.
11:44 vendethiel left
donaldh I'm guessing callframe.line should always return a value ? 11:47
11:47 yqt joined
donaldh Or should Test.pm just be defensive ? 11:53
11:54 woshty left, woshty joined 11:57 pecastro joined 12:01 coffee` joined
moritz doesn't know 12:03
12:04 vendethiel joined, gfldex joined
masak .u ∘ 12:09
yoleaux U+2218 RING OPERATOR [Sm] (∘)
masak .u ⊚
yoleaux U+229A CIRCLED RING OPERATOR [Sm] (⊚)
masak .u ≗ 12:10
yoleaux U+2257 RING EQUAL TO [Sm] (≗)
masak lots of Unicode custom operator potential here... :)
nine_ Unicode operators would be much more fun if they were easier to type
masak nine_: here's a nickel. buy yourself a decent text editor. 12:12
:P
|Tux| gist.github.com/Tux/6a740c1b2a7455361cfd - Unicode OPERATOR's 12:13
masak srsly, though. computers are programmable. if iPhones can be made to show custom keybords, then so can your laptop or desktop.
nine_ masak: is vim decent? 12:14
masak nine_: yes.
nine_: ^K ftw
nine_: Emacs is also decent. there's an input method module called LEIM. 12:15
nine_ niiiiice
|Tux| ~/.XCompose is also decent :P
masak I don't know about all these other newfangled things, though. people are all sublime textmates these days.
moritz ~/.XCompose has the advantage that it works in every application 12:16
masak should ~/.XCompose more, clearly
|Tux| <Multi_key> <C> <equal> : "€" U20a0 # EURO-CURRENCY SIGN
<Multi_key> <c> <equal> : "€" U20a0 # EURO-CURRENCY SIGN
<Multi_key> <equal> <C> : "€" U20a0 # EURO-CURRENCY SIGN
<Multi_key> <equal> <c> : "€" U20a0 # EURO-CURRENCY SIGN
masak :)
|Tux|: but what about lower-case euro? :P
|Tux| I don't ever use it :P 12:17
nine_ Now if only there was an easy way to type in unicode characters to search in vim's digraph list :P
lizmat FROGGS: should I bump NQP_REVISION ?
abraxxa CTRL-K does nothing in my vim
masak nine_: after a while you kind of just pick up the principle of it all
abraxxa: `help digraphs` 12:18
abraxxa: `:help digraphs`
12:20 kjs_ left
nine_ And of course » is just ^K>> :) 12:22
abraxxa masak: thanks, reading
masak nine_: see? you're already picking it up :) 12:23
nine_: similarly, Greek letters all have a '*' in them
moritz because the Greek conquered the stars. 12:24
nine_ Σ FTW :)
masak Cyrillic all have a '=' in them. because the Russians conquered... equality?
|Tux| I always missed the IOTA SUBSCRIPTUM in greek, then I could have written T\N{IOTA SUBSCRIPTUM}o for a new type Toyota: the subscriptum 12:27
Τῳ 12:28
.u Τῳ
yoleaux U+03A4 GREEK CAPITAL LETTER TAU [Lu] (Τ)
U+1FF3 GREEK SMALL LETTER OMEGA WITH YPOGEGRAMMENI [Ll] (ῳ)
12:29 Ugator joined
masak .oO( I don't always type the Greek small letter omega -- but when I do, I type it with ypogegrammeni ) 12:33
smls What's the difference between replies and comments on rt.perl.org? 12:35
moritz replies go to the requestor 12:36
nine_ smls: RT is a system for help desks. Comments are internal (for the team) while replies also go to the customer
smls so which is preferred in rakudo bugs? 12:37
nine_ I'd use reply
Even if the reply does not actually help the requestor, it at least shows some activity. 12:38
masak heh. in all my years with RT, I hadn't bothered learning the difference. :) 12:44
but yeah, I always "Reply", too.
Ven .tell eli-sde github.com/pixie-lang/pixie 12:47
yoleaux Ven: I'll pass your message to eli-sde.
12:47 kjs_ joined
Ven .tell eli-se github.com/pixie-lang/pixie 12:47
yoleaux Ven: I'll pass your message to eli-se. 12:48
12:51 brrt joined 12:56 yqt left 13:10 brrt left, vendethiel left
lizmat FROGGS: FWIW, after an NQP bump, JVM is still distinctly unhappy 13:12
13:13 brrt joined
FROGGS ahh sorry, was distracted 13:17
13:18 vendethiel joined
moritz news.perlfoundation.org/2015/03/tak...eachy.html should we try to get some Perl 6 ideas and mentors in there? 13:21
donaldh lizmat: did you try my last commit?
lizmat donaldh: probably not, is it a pull request ?
nine_ moritz: we definitely should! 13:22
donaldh lizmat: I've commited a change to nqp.
lizmat and I locally raised NQP_REVISION to that
testing with 2015.02-79-g1dfa4ca 13:23
donaldh will test from clean again 13:24
13:28 Ven left
dalek kudo/nom: b8a4665 | lizmat++ | t/01-sanity/22-KEY.t:
Add Set/Bag/Mix(Hashes) sanity tests
 13:28
masak sanity++ lizmat++ 13:29
lizmat re #124023, I wonder whether this should be fixed at the rakudo level, or lower? 13:32
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124023
lizmat aka, automatic \n conversion to \r\n on Win in nqp / Moar ? Or will we do this in sub say ? 13:33
moritz first we should have a consensus how it should look at the Perl 6 level
and then we should think about where to implement it