»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.
00:08 rurban left 00:10 virtualsue left
jnthn is tired... 00:18
Will look some more tomorrow
'night
00:25 ens left, ens joined
hoelzro night 00:28
00:31 rindolf joined 00:32 panchiniak left
lue I still can't install panda at all :( Cannot call 'postcircumfix:<{ }>'; none of these signatures match: 00:33
00:48 xenoterracide left 00:51 thou left 01:00 dmol left 01:03 jnap left 01:08 xenoterracide joined 01:09 rurban joined 01:13 btyler left 01:25 tgg joined, tgg left 01:34 jnap joined 01:37 kbaker joined 01:45 kbaker left 01:52 woosley joined 01:58 jnap left 02:00 lizmat joined, lizmat left 02:05 eternaleye left 02:14 johnmilton left 02:27 logie left 02:41 rurban left 02:50 FROGGS_ joined 02:51 ens is now known as SELL 02:52 Sqirrel_ left, Sqirrel joined 02:54 FROGGS left
japhb timotimo: gist.github.com/timo/544d3d75bd92932e635d has entered the realm of pure unadulterated awesome. Now do it for gen2. ;-) 03:03
hoelzro: gist.github.com/anonymous/8982533 is also very nice to see. Now to get it to 100% OK. :-) 03:04
03:06 SELL is now known as ens 03:11 aindilis left 03:46 woosley1 joined 03:48 woosley left, apejens joined 03:49 nwc10 left, nwc10 joined 03:54 ens is now known as SELL 03:59 rurban joined 04:08 HazMat_ joined
raydiak jnthn: just noticed the backlog about typed arrays from 20 hours ago...thank you for showing us how to do it properly, and I will try not to disseminate any more ignorance :) 04:08
04:09 HazMat_ left
raydiak s/20/19/ 04:12
04:13 xenoterracide left 04:21 woosley1 left 04:22 araujo left, SamuraiJack joined 04:23 eternaleye joined 04:29 dayangkun left
timotimo has to take a break from sleeptrying 04:30
04:35 woosley joined 04:36 SamuraiJack left
TimToady rosettacode.org/wiki/Parsing/Shunti...thm#Perl_6 04:36
of which the most interesting point is that it overrides the meaning of "shift" and "reduce" in an inner scope 04:37
timotimo what were you thinking, overriding keywords just like that! ;) 04:38
TimToady blames it on the flu :) 04:39
04:39 xenoterracide joined
TimToady I suppose I coulda made them multis 04:40
timotimo i was just about to ask for a cool implementation of wrapping a list of True/False up in a hilbert curve and displaying them with the unicode upper/lower half block + full block 04:41
but then i remembered i have to do this in python
lue PL/I, Perl 6, and REXX have the nicest at-a-glance outputs imo
TimToady bows, having considered various options 04:42
04:42 dayangkun joined
timotimo has a sneaking suspicion that 256 buckets per page and 32 different object size buckets makes for a whole lot of untaken space 04:45
i quite hope that that's the main contributor to 100 megabytes of ram for "say 1" in perl6-m 04:47
because that would make for easy tuning :P
05:03 SELL is now known as ens 05:26 [Sno] left 05:27 ponbiki_ is now known as ponbiki 05:33 BenGoldberg left
moritz \o 05:34
timotimo o/
05:36 btyler joined
raydiak morning moritz 05:38
05:47 kaleem joined
raydiak timotimo: it's not python, but as a point of interest, there is hilbert ordering and combining multiple points into single characters in the concurrency branch of Pray 05:55
it's messy for 2 or 3 different reasons, though, so "cool implementation" not included :) 05:56
timotimo en.wikipedia.org/wiki/Hilbert_curve ← i ported xy2d from this article 06:01
what can i say ... it hilberts :D
raydiak timotimo: wouldn't have helped much, then...I ported from the same article...though only the d2xy, now that I think about it 06:03
timotimo i must admit, this is darn pretty
raydiak what are you drawing, anyway? 06:04
timotimo gist.github.com/timo/1f4fa4bc76b270829382
how much of the gen2 pages are used vs allocated
kind of not so nice that it takes up so much space vertically if there's only one page for all the size buckets 06:05
but oh well
raydiak ah, right, the heap analyzer
timotimo i ought to run a longer program that actually punches a few holes in there
raydiak Pray is great for punching holes in things :P 06:06
timotimo perl6-m doesn't do permutations yet :(
r: say [1, 2, 3, 4].permutations()
camelia rakudo-moar 1ce379: OUTPUT«Cannot call 'postcircumfix:<[ ]>'; none of these signatures match:␤:(Any \SELF, int $pos)␤:(Any \SELF, int $pos, :BIND($BIND)!)␤:(Any \SELF, int $pos, Any :SINK($SINK)!, *%other)␤:(Any \SELF, int $pos, Any :delete($delete)!, *%other)␤:(Any \SELF,…» 06:07
..rakudo-parrot 1ce379, rakudo-jvm 1ce379: OUTPUT«1 2 3 4 1 2 4 3 1 3 2 4 1 3 4 2 1 4 2 3 1 4 3 2 2 1 3 4 2 1 4 3 2 3 1 4 2 3 4 1 2 4 1 3 2 4 3 1 3 1 2 4 3 1 4 2 3 2 1 4 3 2 4 1 3 4 1 2 3 4 2 1 4 1 2 3 4 1 3 2 4 2 1 3 4 2 3 1 4 3 1 2 4 3 2 1␤»
timotimo r: say (1, 2, 3, 4).permutations()
camelia rakudo-moar 1ce379: OUTPUT«Cannot call 'postcircumfix:<[ ]>'; none of these signatures match:␤:(Any \SELF, int $pos)␤:(Any \SELF, int $pos, :BIND($BIND)!)␤:(Any \SELF, int $pos, Any :SINK($SINK)!, *%other)␤:(Any \SELF, int $pos, Any :delete($delete)!, *%other)␤:(Any \SELF,…»
..rakudo-parrot 1ce379, rakudo-jvm 1ce379: OUTPUT«1 2 3 4 1 2 4 3 1 3 2 4 1 3 4 2 1 4 2 3 1 4 3 2 2 1 3 4 2 1 4 3 2 3 1 4 2 3 4 1 2 4 1 3 2 4 3 1 3 1 2 4 3 1 4 2 3 2 1 4 3 2 4 1 3 4 1 2 3 4 2 1 4 1 2 3 4 1 3 2 4 2 1 3 4 2 3 1 4 3 1 2 4 3 2 1␤»
06:12 aindilis joined 06:14 thou joined 06:17 woosley left
timotimo ████████████████ 06:18
██████▄█████████
████████████████
look!
a hole! :D
Pleiades` I was going to interrupt it, but idk if that would be considered rude here
timotimo (and 1426 empty pages) (freelist with 1 entries)
that doesn't seem right?! o_O
JimmyZ moarvm?
timotimo yup
those pages all seem to have been allocated. huh. 06:21
hm, maybe with null pointers though 06:22
06:23 [Sno] joined
timotimo holy jesus, this is a HUGE free list 06:26
06:26 cognominal__ left
timotimo or i'm just infinilooping 06:27
yup. 06:28
cur_page doesn't get bumped 06:34
wow. 06:37
this size bucket gets quite some use out of it.
gist.github.com/timo/1f4fa4bc76b270829382 - what sayeth you? 06:49
06:57 FROGGS_ left 07:01 darutoko joined
moritz \o from a car 07:03
timotimo: did you mean "what sayeth thou?" :-)
high latency is annoying 07:04
07:04 kaleem left 07:27 fhelmberger joined 07:34 btyler left 07:48 woosley joined 07:51 virtualsue joined 07:52 FROGGS joined 07:59 denis_boyun joined 08:02 zakharyas joined, pdcawley joined
timotimo resumes his sleeps 08:04
08:05 cognominal joined, thou left
tadzik FROGGS: pong 08:10
08:11 virtualsue left
FROGGS tadzik: are the File::Find tests supposed to pass on windows boxes? 08:11
tadzik FROGGS: I think so 08:12
FROGGS because these are hardcoded to forward slashes :/
tadzik oh 08:13
FROGGS I am just considering making another test file for windows only, and make the current one unix only 08:14
tadzik I'd rather have cross-platform paths everywhere 08:15
if they're too much of a hassle, let's think how can we make them less hassl
FROGGS what do you mean by cross platform paths?
you are supposed to get a lib\File back on windows instead of lib/File, no? 08:16
I mean, windows does understand forward slashes nowadays... but... 08:17
08:25 dmol joined, kaleem joined 08:26 SamuraiJack joined, mtk left
tadzik I mean, stop using "path strings", use proper path objects everywhere necessary 08:28
08:30 lizmat joined
FROGGS well, then we should just adjust the tests to not compare against strings, no? 08:30
tadzik yes 08:31
FROGGS r: say so "foo/bar".IO.path ~~ "foo\\bar".IO.path
camelia rakudo-parrot 1ce379, rakudo-jvm 1ce379, rakudo-moar 1ce379: OUTPUT«False␤»
FROGGS :o(
08:32 pecastro left
tadzik perhaps we need something like github.com/odyniec/File-PlainPath :) 08:34
otoh, we may just want it in core
FROGGS I'd propose that IO::Path.ACCEPTS should compare the volume/directory/file components without caring about the path separator 08:35
tadzik sounds good to me
08:39 Sqirrel left 08:40 kshannon_ left, arnsholt joined
arnsholt o/ 08:40
08:40 kshannon joined
FROGGS moin arnsholt 08:42
08:43 kaleem left 09:02 woosley left
arnsholt Ooooh... it's working =D 09:11
(Sort of)
masak good morning, #perl6 09:12
Perl 6 day today! \o/
finally!
nwc10 what awesomeness are you going to cook up for us? 09:13
(or for yourself)
or is it a surprise?
arnsholt I've got callbacks returning void working to the point where I need to implement the delegation back to the NQP/Perl 6 callback function 09:14
masak nwc10: 50% p6cc/t4, 50% secret project :)
arnsholt Something wrong with the code generated for non-void return callbacks though 09:15
09:18 SamuraiJack left 09:23 woosley joined
arnsholt JVM segfault! \o/ 09:23
That usually means my code is at least doing *something* =D 09:24
09:24 Alina-malina left 09:29 aindilis left
FROGGS \o/ 09:29
progress++
err
arnsholt++
:o)
09:30 denis_boyun left
tadzik :) 09:30
09:35 denis_boyun joined, Alina-malina joined, Alina-malina left, Alina-malina joined
mathw masak: secret project! Nooooo 09:38
mathw is too impatient for secret projects 09:39
09:39 denis_boyun left
masak mathw: ok, allow me to be a bit more specific. 09:40
mathw: it's a tool for $work, that I'm writing on my spare time because I want to spring it on my colleagues.
09:41 darutoko- joined
mathw I hope they appreciate it then :) 09:43
09:43 darutoko left
masak I'm probably gonna blog about it, too. 09:44
09:46 Alina-malina left, Alina-malina joined
mathw I wonder if I can get a job where I don't have to do .NET web development anymore 09:47
masak mathw: we're hiring. 09:49
09:49 nsh_ joined
masak (and we're a really cool company, with two Perl 6 core devs in it.) 09:49
mathw And you're in Sweden, which is slightly inconvenient 09:51
09:51 nsh left
mathw (not for you, obviously) 09:51
mathw wonders, does being in Sweden make you more cool? Probably. 09:53
09:54 xenoterracide left, pdcawley left, pecastro joined 09:56 denisboyun joined 09:58 rurban left
masak from the perspective of Sweden, Sweden is not that cool. unless you mean literally cool. 09:58
09:59 kivutar joined, woosley left, pdcawley joined 10:00 denis_boyun joined
nwc10 from the perspsective of outside Sweden, Sweden generally seems to be cool. 10:01
10:01 denisboyun left
nwc10 although it's not actually clear what or why. :-) 10:03
cosimo_ well, ABBA to start with
10:03 lizmat left
nwc10 that's almost like saying that Austria is cool because it has Franz Joseph, Sissy and Mozart 10:04
which they are very big on.
10:04 denis_boyun left
nwc10 but are a little bit history 10:04
cosimo_ roxette, a nice princess, cheap meat, ikea kjøttboller
nwc10 and very dead. Unlike the (former) members of Abba
10:04 denisboyun joined
cosimo_ not to mention karolinska institute and chalmers 10:05
FROGGS and salmiak "sweets"? 10:06
masak hehe, "kjøttboller" is the Danish spelling :P
also, salmiak totally rocks.
FROGGS >.<
like roxette?
FROGGS hates roxette 10:07
cosimo_ unnskyld
masak ;)
FROGGS: I've never had a big crush on Roxette. they're OK, I guess. but nothing special.
FROGGS the problem is that I was forced to hear it in my childhood 10:08
the radio stations played it like every 30mins
10:08 denis_boyun joined
FROGGS sleeping in my car *shudder* 10:08
10:09 denisboyun left
jnthn morning...ish... o/ 10:09
FROGGS morning jnthnish 10:10
(you are not yourself when you are hungry)
masak FROGGS: yeah... "sleeping in my car" is cringeful.
jnthn Well, it's more coffee that's needed...
FROGGS jnthn: that is the right attitude :o) 10:12
masak: exactly ó.ò
10:13 denis_boyun left
jnthn Wow, those lyrics are...uh...to the point... 10:14
masak and the video leaves nothing to the imagination. 10:15
10:19 denisboyun joined, Sqirrel joined 10:24 denisboyun left, nsh_ left, nsh_ joined 10:25 nsh_ is now known as nsh, SamuraiJack joined 10:36 grondilu joined
mathw masak: I suspect if I actually lived in Sweden I'd quickly find out that it's not so cool as it seems from the outside. 10:37
masak let's just say I seldom hear people exclaim "oh, it's so cool to live here in Sweden!" :P 10:38
except perhaps if their basis of comparison is Syria or Somalia.
mathw lots of places are better than Syria or Somalia at the moment 10:40
Even the parts of England currently under water
masak yeah. 10:41
mathw despite what our gutter press would have you believe
jnthn Ugh...the English press... 10:42
mathw quite 10:43
not my favourite part of living here
jnthn Yeah, that's one thing I certainly don't miss.
mathw I generally attempt to ignore it, but that's not always possible
And some things have to be actively countered 10:44
10:49 kaleem joined 10:50 kivutar left
tadzik mathw: .NET? Around here, it's practically impossible to find something other than C++ :) 10:55
masak tadzik: that's because Sweden has sold out to Microsoft and is their obedient lapdog. 11:02
tadzik :)
masak tadzik: similarly, in Norway it's mostly Java.
tadzik I'd imagine Norway would sell _everything_ for some Sun...
otoh, they seem unimpressed when the sun is there :) 11:03
masak when the sun is there, it's time to Gå På Tur. 11:05
mathw Nottingham seems to be .NET land, but that might be my perceptions because of what recruiters see on my LinkedIn profile 11:07
there is a fair bit of PHP around, but I'm not that desperate
11:08 sqirrel_ joined
tadzik I always thought England is a magical Perl wonderland, but maybe that's just London 11:11
jnthn Not to mention that PHP pretty much promises you'll be doing web dev...
masak getting some warnings when I run Rakudo's Configure.pl: gist.github.com/masak/8999395 11:13
haven't looked at the cause, but I think warnings from a Configure.pl script are suspect by default.
it's $nqp_config{'nqp::version'} that's undefined. 11:14
and then $nqp_config{"moar::dll"}
mathw jnthn: yeah that's not even why I avoid it 11:15
tadzik: there's a lot of Perl in London. Really seriously a lot.
masak yeah, turns out I couldn't `make` either, because it's missing NQP.
needed to add --gen-nqp to the Configure.pl script. 11:16
...now I'm getting an error instead :(
jnthn masak: What're you trying to get?
masak help! gist.github.com/masak/8999434
masak tries with a clean Rakudo repo 11:17
jnthn I have to follow a link??!
masak: You need to --gen-moar also
masak: Or specify a --prefix=...
masak: To where you installed Moar
masak oh ffs
used to be the Configure.pl script told me what was missing.
now it just errors out? :(
jnthn That is unfortunate 11:18
otoh, README does give you the exact thing you need... :)
masak submits rakudobug
jnthn But nobody reads that, of course :P
masak I'm a core dev, I don't read the README! :P
nwc10 that's kind of rule 0 of programming - no one reads the docs 11:19
and some people have a rule 1, of not reading the error message either
masak anyway, I think this is orthogonal.
nwc10 but those folks should be issued with an Etch-A-Sketch
masak I *did* read the error message, and it sucked.
mathw LTA error message is LTA
nwc10 yes, that's a legitimate bug
or, I agree, that's a bug.
jnthn It did say: 11:21
===SORRY!===
No suitable MoarVM (moar executable) found using the --prefix
masak yeah. that's not enough.
jnthn OK, what should it have said?
"(Maybe you want --gen-moar?)"
masak I mean, you're *intended* to use Configure.pl in a certain way, and the error message doesn't give enough hints as to how.
jnthn: yes, something like that.
whatever message would have saved me a trip to #perl6 for help would have been fine ;)
mathw Welcome to the Rakudo Perl 6 configuration system 11:22
Press 1 to gen-moar
Press 2 to gen-parrot
masak ;)
tadzik :)
mathw Press 3 if you're really patient and want to gen-jvm
jnthn masak: Like, "(Maybe you should RTFM?)" :D
FROGGS Press * to talk to an operator 11:23
( Hi, Mr. Tight And is speaking... )
masak jnthn: I'm all for reading the docs too, it's not that.
tadzik yeah, I'd just keep pressing 0 until I can talk to a human :)
masak jnthn: but a decent bet is that 'perl Configure.pl && make && make install' works in a project.
mathw I've actually written those telephone menu systems
jnthn mathw: Those systems are even more fun when they're not in your first language :) 11:24
masak jnthn: and the current Configure.pl breaks that expectation, without giving an error message that guides the user right.
mathw Designing the options and flow is really hard
FROGGS throws a num pad at mathw
masak jnthn: think of it as saving or losing all the other potential Rakudo users that happened to take the same path as I did.
jnthn: (and who don't know to go to #perl6 for help, or check the README)
redundancy is a good thing here. 11:25
mathw FROGGS: it's highly unlikely you've used one of mine, unless you're a Special Constable or you are unlucky enough to work for companies who use a certain automated sickness notification system 11:26
dalek p: 1658541 | jnthn++ | Configure.pl:
Give a hint about --gen-moar when no moar is found
mathw and who weren't particularly fond of listening to feedback on their design, so never listened to us saying 'you know, the chance any given company has two employees with the same date of birth is pretty high' 11:27
jnthn masak: Hopefully 1658541 helps a little
masak jnthn++ 11:28
ok, resolving the rakudobug.
jnthn masak: Does a straight "perl Configure.pl" with no options say something sane, ooc?
mathw that might be the first time I've seen an all-numeric git hash
masak jnthn: dunno, didn't try that. 11:29
11:29 denisboyun joined
jnthn Hm 11:29
masak jnthn: lemme try with a clean rakudo.
jnthn C:\consulting\rakudo>perl Configure.pl
No suitable nqp executables found! Please specify some --backends, or a --prefix that contains nqp-{p,j,m} executables
I suspect that one can be more awesome. 11:30
masak aye.
because I did specify --backends=moar
and when that didn't work, I thought to specify --gen-nqp 11:31
and when *that* didn't work, I lost my patience and started kvetching here. :)
jnthn ugh, yeah, just --backends=moar has a bad failure mode.
masak aye. it has warnings about undef values. 11:32
&
11:35 sisar joined
sisar o/ 11:35
MoarVM build fail : gist.github.com/Siddhant/8999646 Thoughts ?
jnthn Well, that's a new one... 11:36
sisar I'm on Cygwin, 32-bit
jnthn sisar: Platform?
11:36 darutoko joined
jnthn Oh. 11:36
sisar oh wait, 32-bit ain't supported ?
jnthn 32-bit seems to have problems in some places, but not all. 32-bit Linux x86 seems to work out fine, for example 11:37
Looking at the error, it's actually libuv that seems to run into trouble.
sisar: github.com/joyent/libuv/issues/845 is the libuv issue 11:38
sisar looks
jnthn Which seems to boil down to, "Cygwin doesn't really provide pthreads" 11:39
11:39 darutoko- left
sisar jnthn: yeah. Well, that's disappointing. -( 11:41
jnthn sisar: Yeah. I guess the best hope is that somebody will want Node.js on Cygwin enough to do the libuv port in aid of that... 11:46
sisar: For building it there, anyway
sisar: I suspect the best way to get Moar usable from Cygwin will be to build a native Windows binary and just make it available in the path, so it can be used from cygwin. 11:47
sisar jnthn: hmm, I can try that.
jnthn: Well, i get a Parrot build fail too ! gist.github.com/Siddhant/8999784 11:49
moritz \o 11:55
12:07 Adriaaan joined
timotimo o\ 12:15
12:19 dayangkun left 12:21 dayangkun joined 12:23 felher joined
masak /o\ 12:34
moritz |o| 12:36
timotimo lol 12:37
FROGGS \o\ 12:41
timotimo meeeeh, reading out the size of every single bucket in every single page is likely going to take minutes if not hours 12:42
i should somewhat randomly sample.
jnthn In gen2? 12:44
timotimo yes
jnthn The pages are sized?
So all the objects on a given page have the same size
timotimo the size bucket of 32 to 39 has 1423 completely filled pages
oh? 12:45
jnthn Oh, yeah
You're right, they may vary a bit
I thought it was organizzed so the sizes would be more like 33..40 though?
So each 8 byte increment gets non-wasting... 12:46
timotimo oh?
well, that may very well be
i didn't check especially thoroughly
that makes a whole lot more sense than having 32-40
perigrin www.yapcna.org/yn2014/talks/tag/perl6 # just gonna mention the deadline is tomorrow at midnight EST 12:47
util++
timotimo yes, you're right
jnthn timotimo: OK, good.
timotimo should i still go ahead and do the sampling? 12:48
jnthn timotimo: Well, good in so far as "I implemented it sanely". Not so good as in, "no easy win here" :)
timotimo: I think more useful is looking at fragmentation.
timotimo right
yeah, i'm doing that
i see almost zero fragmentation in my example program, which is not especially carefully designed
do you have a suggestion for something that would fragment heavily?
12:49 benabik left
jnthn Yeah...something that makes loads of objects of a certain size, then stops making them and does other stuff, may well cause it. 12:49
At the moment it doesn't make any attempt to say, "oh, this entire page is on the freelist, let's free it" 12:50
But what yo're doing can probably tell us how pressing it is to implement that. 12:51
timotimo yes
<3
i don't have a metric for how many pages "in between" are empty yet 12:52
my stuff already helped me find out that the cur_page value isn't being updated at all 13:00
FROGGS yeah 13:01
analysis++
timotimo a truly tremendous oversight! :P
jnthn timotimo: this: 13:02
/* The current page number that we're allocating in. */
MVMuint32 cur_page;
That appears to be completely unused.
timotimo i use it! :)
in my analysis
jnthn Yeah but...the rest of the VM doesn't :) 13:03
timotimo aaw
jnthn I suspect it's fossil
timotimo well, we can save 32 bits * 32 buckets if we get rid of it!
jnthn Every little helps :) 13:04
dalek kudo/nom: c50a2e8 | (Tobias Leich)++ | tools/build/Makefile-Moar.in:
do not touch ./perl6-m, only the one in install/

Because touching the runner in the rakudo build directory means triggering a rebuild of lib.pm, Test.pm and Pod/to/Text.pm.
jnthn timotimo: Feel free to remove it, anyways, given it seems unrequired.
timotimo: If we don't find something in the free list, then it's clear that the "current" page to try allocating on is the final one. 13:05
So num_pages is sufficient.
timotimo i suppose i can make it work without that attribute somewhat easily
but first, i have other stuff to do with this 13:06
FROGGS perl6-m -e 'say IO::Path::Win32.new("foo\\bar") ~~ "foo/bar/".IO.path' 13:07
True
p: say IO::Path::Win32.new("foo\\bar") ~~ "foo/bar/".IO.path
camelia rakudo-parrot 1ce379: OUTPUT«False␤»
timotimo i shall sample the REPRs there as well, aye? 13:08
jnthn timotimo: aye, that'll tell us what's in there
moritz FROGGS: .path is enough, no need for .IO 13:14
FROGGS moritz: but it feels better to use .IO somehow :o) 13:15
moritz FROGGS: but eventually it'll be wrong
FROGGS: the problem with a path-based API is that it's prone to race conditions 13:16
FROGGS I am pushing the added IO::Path.ACCEPTS when no p-spectests fail
moritz FROGGS: so the long term plan is to use .IO to offer a file-handle based API
FROGGS ahh, I see
moritz which won't be as convenient, but will be safer for some scenarios
hoelzro r: 'ш'.encode('latin-1').decode('utf-8') 13:22
camelia ( no output )
..rakudo-parrot 1ce379: OUTPUT«Lossy conversion to single byte encoding␤ in method encode at gen/parrot/CORE.setting:6281␤ in block at /tmp/tmpfile:1␤␤»
hoelzro r: say 'ш'.encode('latin-1').decode('utf-8')
camelia rakudo-jvm 1ce379, rakudo-moar 1ce379: OUTPUT«?␤»
..rakudo-parrot 1ce379: OUTPUT«Lossy conversion to single byte encoding␤ in method encode at gen/parrot/CORE.setting:6281␤ in block at /tmp/tmpfile:1␤␤»
hoelzro what *should* that output?
jnthn I prefer the JVM/Moar answer 13:23
hoelzro I think that a :strict option would be nice
jnthn Yeah, same
moritz no, :strict shoudl be the default 13:24
there could be a :lax option which uses a replacement character
hoelzro r: "\x[82]".encode('latin-1').decode('utf-8')
camelia ( no output )
..rakudo-parrot 1ce379: OUTPUT«Malformed UTF-8 string␤␤ in method decode at gen/parrot/CORE.setting:5117␤ in block at /tmp/tmpfile:1␤␤»
hoelzro moritz: I agree that it should be the deafult
*default
r: say "\x[82]".encode('latin-1').decode('utf-8')
camelia rakudo-moar 1ce379: OUTPUT«?␤»
..rakudo-jvm 1ce379: OUTPUT«�␤»
..rakudo-parrot 1ce379: OUTPUT«Malformed UTF-8 string␤␤ in method decode at gen/parrot/CORE.setting:5117␤ in block at /tmp/tmpfile:1␤␤»
jnthn moritz, hoelzro: That's a fine default until the day you meet real world data that sucks... :/
hoelzro ok, now *that's* interesting
moritz jnthn: then you start to set the :lax flag :-) 13:25
hoelzro that's what the problem was with URI
I'm looking to create an NQP test for this
but I don't know what the results *should* be
jnthn hoelzro: The "malformed UTF-8 string" error doesn't make a lot of sense there
p: say "\x[82]".encode('latin-1')
camelia rakudo-parrot 1ce379: OUTPUT«Buf:0x<82>␤»
hoelzro yeah, I think Parrot's doing something stupid there 13:26
jnthn Oh
No, it's right to complain I think.
82 is not a valid UTF-8 sequence.
Well
FROGGS everything > 0x80 is invalid, yes
jnthn But is it valid in latin-1?
13:26 Sqirrel left
FROGGS p: say "\x[82]" 13:27
camelia rakudo-parrot 1ce379: OUTPUT«␤»
moritz isn'T every byte squence valid Latin-1?
jnthn moritz: Well, it's more about if Unicode codepoint 0x82 has a mapping in latin-1
"\x[82]" is a string so we're clearly talking Unicode codepoint there.
moritz I kinda thought the first 256 Codepoints map directly to the same byte values in Latin-1 13:28
seems I'm wrong though; Latin-1 has holes
timotimo the first 127 surely do :P
jnthn moritz: No, the mapping ain't that simple 13:29
hoelzro I think that 0x82 encoding to 0x82 in latin-1 is the bad behavior, right?
jnthn U+0082 is Break Permitted Here in Unicode.
moritz it seems that windows-1252 (a latin-1 superset) is closer to what I though 13:30
jnthn And looking at en.wikipedia.org/wiki/ISO/IEC_8859-1 suggests that this character doesn't have a latin-1 representation. 13:31
hoelzro: So yes, codepoint 0x82 encoding to latin-1 byte 0x82 seems wrong
hoelzro ok, good 13:33
jnthn I agree we should support strict and lax modes for encoding/decoding 13:34
I'll leave TimToady++ to decide the default :)
nwc10 en.wikipedia.org/wiki/ISO/IEC_8859-1 isn't totally helpful, as it seems to omit the C0 and C1 control codes from its table. 13:35
jnthn nwc10: C0 and C1? It defines those as À and Á 13:36
nwc10 no, these: en.wikipedia.org/wiki/C0_and_C1_control_codes 13:38
13:39 skids left
jnthn nwc10: Oh. :/ 13:39
nwc10: So if you include those, 0x82 is valid...
TimToady I count latin-1 as the first 256 codepoints of Unicode, and I don't really seem much problem in doing so 13:42
*see 13:43
jnthn TimToady: So...all the clever mapping code in github.com/MoarVM/MoarVM/blob/mast...s/latin1.c is bogus?
"clever" :) 13:44
dalek kudo/nom: 6a6a5bd | (Tobias Leich)++ | src/core/IO.pm:
add IO::Path.ACCEPTS

A path will match another path if their volume/directories/file portion matches.
13:45 Khisanth left
TimToady yeah, I think it's bogus 13:45
jnthn That seems to mean we get to delete code :) 13:46
timotimo \o/
jnthn en.wikipedia.org/wiki/Latin-1_Suppl...ode_block) seems to agree also
13:50 fhelmberger left
nwc10 jnthn: is that mapping Windows 1252? 13:50
13:50 FROGGS left
jnthn nwc10: Oh...hmm 13:52
nwc10: That is entirely possible.
nwc10: Which I guess is fine to support but...it should not be claiming to be latin-1... 13:53
13:54 sqirrel_ left
nwc10 unless of course you are HTML 5, IIRC 13:54
TimToady
.oO(ISO-8859-MS)
timotimo m) 13:55
mathw yeah HTML 5 says that if something declares latin-1 parse it as windows-1252 13:57
I think I'd've been tempted to say if it claims to be anything that's not unicode, insult them, but I suppose backwards compatibility might have suffered
TimToady ick 13:58
13:58 Khisanth joined
TimToady and if it doesn't claim to be anything, it's probably shift-jis :P 13:59
nwc10 I find it utterly ick too. 14:00
TimToady so if it's ambiguous, perhaps we should simply disallow 'latin-1', and require clarity
it's either that, or swing our Bat of Cultural Imperialism one way or the other 14:01
nwc10 actually, it's worse. HTML 5 "requires that documents advertised as ISO-8859-1 actually be parsed with the Windows-1252 encoding."
TimToady AGHHH! 14:02
nwc10 not "Latin 1", but a declaratio of a *real* ISO standard.
yes. urge to kill.
TimToady well, maybe we've lost that battle, and need ascii-256 or so 14:03
TimToady wonders which circle in hell is reserved for people who write standards... 14:04
nwc10 hopefully not a round one.
moritz probably depends on the standard 14:05
moritz doesn't wish hell for the guy who standardized JSON
TimToady wishes JSON allowed trailing commas 14:06
timotimo too
moritz too, but that's not a hell-banning reason 14:07
sjn maybe the upper levels of heck then? :)
TimToady most of us at least deserve some penance in purgatory :) 14:08
sjn has always had a thing for limbo 14:09
TimToady was never very good at it
moritz is the purgatory still in the newe testament? 14:10
TimToady not in the protestant canon 14:11
dunno about the NT apocrypha 14:12
TimToady always thought of it as an accretion that went along with indulgences 14:13
my basic feeling is that, if HTML-5 wants to lie to itself, that HTML-5's problem, and not ours (unless we're writing an HTML-5 module) 14:16
so unless the whole industry buys the lie, we should distinguish iso-8859-1 from windows-1252 14:17
(and latin-1 with it) 14:18
mathw There's a certain rightness to saying "Well it said it was ISO-8859-1, if it was lying that's not our problem"
TimToady then there's all the iso-8859-1 that's really iso-8859-15 14:20
arnsholt jnthn: Will the JVM invoke ops DTRT if I use an ARG_OBJ (with corresponding Integer object) to invoke a function that takes a native int? 14:22
masak sjn: I've always had a thing for limbo, too! 14:23
a place to chill.
I imagine if I brought a few books...
jnthn arnsholt: As in, Java Integer? No 14:24
timotimo masak: dresdencodak.com/2010/05/05/a-degen...-for-hell/ <- :)
masak timotimo: that must have been where I got that from, yes :) 14:25
arnsholt Darn. That'll complicate the callback logic a bit, but nothing more code can't solve =) 14:26
14:26 FROGGS[mobile] joined 14:27 kaleem left 14:30 jnap joined 14:32 lowpro30 joined 14:36 SamuraiJack left 14:39 guru joined, guru is now known as Guest63253, sftp left, Guest63253 is now known as ajr_
mathw arnsholt: nothing can't be solved by more layers of abstraction! Nothing! 14:41
masak is there an RT ticket for lines() not being lazy, or for loops not being lazy or something in that area not being lazy? 14:42
14:42 sftp joined
moritz is pretty sure tehre is at least one 14:42
masak can't demonstrate it on the evalbot, I think.
but it stands out as a rather big, long-standing unsolved annoyance.
masak tries making it a while loop 14:43
14:43 lowpro30_ joined
masak yup, that's a feasible workaround. 14:43
replace 'for $*IN.lines { ... }' with 'while defined( $_ = $*IN.get ) { ... }' 14:44
masak goes hunting for the RT ticket
rt.perl.org/Ticket/Display.html?id=112716
filed in May 2012. 14:45
arnsholt mathw: Apart from having too many abstraction layers =)
mathw didn't I see something about being able to say "eager for" recently, or was I imagining it?
arnsholt: nonsense! you can abstract away from that!
arnsholt =D
moritz mathw: there's a difference between for being eager, and for eagerly evaluating its input list before starting to iterate 14:46
masak oh, so it's the workahead that causes the problem.
14:47 lowpro30 left
masak and pmichaud was working on a fix two years ago, two days after the bug was filed. 14:47
wonder what happened to that fix...
mathw moritz: oh yes, of course. 14:48
moritz: I have a cold that's rapidly getting worse, my brain's not working very well
masak mathw: no, it's just that these things are tricky :)
moritz tricky to implement, too 14:49
dalek kudo-star-daily: 31275d4 | coke++ | log/ (5 files):
today (automated commit)
 14:55
kudo-star-daily: 30943a2 | coke++ | log/ (5 files):
today (automated commit)
rl6-roast-data: 0d6a615 | coke++ | / (3 files):
today (automated commit)
14:56 molaf joined 14:57 erkan left, erkan joined
masak [Coke]++ # hero 14:57
TimToady I suppose the first 256 codepoints should be called "UCS-1" :)
[Coke] got to see Alton Brown in concert last night. Whee! 15:00
TimToady pity you couldn't hear too
15:02 xinming_ joined 15:04 thou joined 15:05 xinming left, pdcawley left 15:06 btyler joined 15:08 kbaker joined 15:14 kaare_ left 15:15 skids joined
[Coke] thankfully he had a projection screen with words most of the time, so you could follow along. 15:16
PerlJam [Coke]: What is an Alton Brown concert like? 15:18
PerlJam imagines a live version of his Good Eats show
15:18 logie joined 15:20 lowpro30 joined
PerlJam never mind, google gave me a better idea of it :) 15:21
15:24 lowpro30_ left
[Coke] PerlJam: that's a pretty good first approximation, though. ;) 15:30
we have no implementations spec testing clean. 15:37
btyler masak: if you've got perl6 time today, perhaps consider taking a look at hoelzro++'s pull against ufo to support multiple backends: github.com/masak/ufo/pull/10 15:39
15:42 daniel-s_ joined 15:43 arnsholt left
raydiak morning all 15:45
jnthn o/ raydiak 15:47
raydiak \o jnthn...doing anything incredible today? 15:48
jnthn raydiak: Lots of $dayjob bits right now, but I'll get to Perl 6 later on and continue with the MoarVM IO work. :) 15:49
15:50 lizmat joined
raydiak :) 15:51
15:53 molaf left
masak btyler: ok, opening it in a tab. might get to it, but no guarantees. 16:00
16:02 daniel-s_ left, Alula left 16:03 Alula joined
masak hoelzro: interesting use of both a class and a group of roles. 16:06
16:06 sftp left
hoelzro thanks 16:06
16:08 denisboyun left 16:09 sftp joined 16:10 dmol left 16:12 dmol joined
jnthn hoelzro: Well, you should check if it's "may you live in interesting times" interesting before saying thanks :P 16:13
hoelzro jnthn: ah, good point =)
masak: is it that kind of interesting?
I'm always looking for critiques on my Perl 6 style
16:17 treehug88 joined 16:18 tgt joined
masak hoelzro: it has more moving parts than I would've had, but since it's an improvement I'm going to merge it in. 16:20
hoelzro masak: cool! how would you have done it?
masak hoelzro: and if I ever get around to adding my improvements, I'll link you the commit ;)
hoelzro =)
masak hoelzro: I would have used a sub instead of a class with a method in it. and a hash of hashes instead of three roles mixed in at runtime. 16:21
dalek o: 2a37cef | (Rob Hoelz)++ | bin/ufo:
Move backend-specific stuff into roles
o: 82067db | (Rob Hoelz)++ | bin/ufo:
Use $*EXECUTABLE_NAME instead of perl6
o: c0f2741 | (Rob Hoelz)++ | bin/ufo:
Move away from Parrot-specific language
hoelzro ah
dalek o: dc80392 | (Rob Hoelz)++ | bin/ufo:
Flesh out JVM role
o: 05bd121 | (Rob Hoelz)++ | bin/ufo:
Strip out POD before looking for dependency modules
masak hoelzro: pull request merged.
o: a4bd3f4 | (Carl Mäsak)++ | bin/ufo:
Merge pull request #10 from hoelzro/master

Make ufo work with non-Parrot backends
hoelzro thank you
masak sorry for flooding #perl6, didn't think of that ;)
16:23 [Sno] left
raydiak is there a way to merge things without flooding? 16:23
jnthn Do it at the command line and --squash, if you don't care for history... ;) 16:25
raydiak ew
16:25 mtk joined
PerlJam why ew? 16:26
raydiak makes thing like bisecting difficult if not impossible
unless I'm confused
PerlJam He did qualify it with "if you don't care for history" :)
flussence Figure out what the threshold for dalek's "heuristic branch merge" message is, and cross that... 16:27
raydiak indeed he did
16:27 woolfy joined
PerlJam
.oO( I had to make 23 minor commits so that dalek wouldn't report the 7 "real" commits on channel )
 16:28
raydiak was just wondering...still figuring out how things work around here (and got dalek kicked last time I merged something) 16:29
timotimo haha 16:30
yeah, dalek tends to do that :)
jnthn raydiak: Thing is, if you do branches to break things for a while, *not* squashing at merge makes bisecting harder. And unfortunately, there's no --first-parent option to bisect... 16:31
raydiak ah, I see what you mean 16:32
raydiak will start squashing simple stuff like docs work 16:34
16:35 colomon left
masak there should totally be a --first-parent option to bisect. 16:36
16:37 tgt left
jnthn Yeah... 16:39
dalek p: 1235a65 | (Elizabeth Mattijsen)++ | src/QRegex/Cursor.nqp:
Introduce .made in favour of .ast, but keep .ast for now
p: fb4ffa9 | (Elizabeth Mattijsen)++ | src/ (4 files):
Change .ast to .made where it seemed appropriate
16:39 ajr_ left
jnthn wonders how hard it'd be to write that patch... 16:39
16:40 guru joined, guru is now known as Guest20329
[Coke] is .ast deprecated? 16:43
16:44 xenoterracide joined, zakharyas left
PerlJam I thought .ast was staying but now there's a .made synonym 16:45
TimToady I don't see much point in keeping both
jnthn My muscle memory :P
But tbh, that only means we keep it in NQP :) 16:46
There's a lot of code out there using .ast already, though. 16:47
16:47 nsh left
PerlJam .ast really makes sense when you're building a compiler ;) 16:48
16:48 Guest20329 is now known as ajr_
jnthn Yeah. 16:50
16:50 woolfy left 16:51 lizmat left
dalek o: d019a08 | masak++ | bin/ufo:
[bin/ufo] turn class/method into nested sub

Limits the scope of this subroutine and shows where it belongs.
Also put the lexically scoped regexes into the sub, since they're only used from in there.		 16:53
o: 0b3a26e | masak++ | bin/ufo:
[bin/ufo] replace mixins with hash and attrs

Prefer the order of the declarations of this variant. In the previous version, it wasn't clear where the $.target and $.extension methods were coming from.
Also, a net removal of 13 lines of code. Neat!
masak hoelzro: ^^^
anyway, hoelzro++ for contributing the multi-backend functionality.
oh!
one more thing I meant to do.
16:59 lizmat joined, woolfy joined 17:00 nsh joined, nsh left, nsh joined
dalek o: a73a527 | masak++ | bin/ufo:
[bin/ufo] give an error if running on unknown VM

Also replace the 'constant' by 'my' scope modifiers, as the former doesn't seem to work at all in this case. So the check is already being useful!		 17:01
17:03 Zaydon joined
hoelzro masak: thanks for pointing that out! 17:07
masak my pleasure. thanks again for the PR. 17:08
17:11 tgt joined
moritz masak++ # p6 coding again 17:11
17:13 arnsholt joined, bluescreen10 joined 17:14 tgt_ joined 17:15 virtualsue joined, integral joined, integral left, integral joined 17:16 tgt left
arnsholt jnthn: Another JVM callsite question. I assume that for something with signature (int, num, object), I'll make a callsite {ARG_INT, ARG_NUM, ARG_OBJ} but the object array passed to the invoke op will contain SMOs for all arguments. That right? 17:17
17:17 benabik joined 17:19 tgt_ left
jnthn arnsholt: No, a Java Integer and Float for the first two 17:19
arnsholt: In the good cases we can optimize away that boxing.
arnsholt Ah, neat. Less work for me 17:20
17:20 tgt joined
arnsholt jnthn: Hmm, does it have to be Integer/Float, or are Byte, Short, Long and Double accepted too? 17:23
jnthn Hmmm
I think it actually wants Double rather than Float...
I guess it casts to unbox
I don't know exactly what happens if you (long) a Byte
17:28 virtualsue left
arnsholt Right, and maybe Long rather than Integer? 17:29
That'd make sense, since we use longs and doubles for numeric types internally
timotimo i long for a good byte of food right now 17:30
jnthn arnsholt: Yds, sorry, meant Long. 17:31
17:31 lowpro30_ joined 17:35 lowpro30 left
dalek p: 15869c8 | (Elizabeth Mattijsen)++ | src/ (12 files):
"past" is now a thing of the ast
 17:35
17:43 bluescreen100 joined 17:44 denisboyun joined 17:46 bluescreen10 left, tgt left 17:47 woolfy left 17:49 lizmat left 18:00 denisboyun left 18:02 SamuraiJack joined 18:10 tgt joined 18:20 zakharyas joined 18:21 FROGGS[mobile] left 18:22 FROGGS[mobile] joined 18:23 pecastro left 18:24 [Sno] joined 18:31 nsh left 18:43 ajr_ left 18:44 guru joined 18:45 guru is now known as ajr_ 18:47 darutoko left 18:53 tgt left, tgt joined 19:01 benabik left 19:02 pmurias joined 19:04 benabik joined 19:05 treehug88 left 19:06 treehug88 joined 19:08 ajr_ left 19:14 tgt left 19:19 SamuraiJack left 19:27 FROGGS[mobile] left 19:30 FROGGS[mobile] joined 19:41 tgt joined, guru joined 19:42 guru is now known as ajr_ 20:01 jnap left 20:12 bluescreen__ joined 20:14 [particle]1 left 20:15 bluescreen100 left 20:17 bluescreen__ left 20:21 rylinaux joined 20:24 kaare_ joined, [particle] joined, FROGGS joined 20:25 telex left 20:26 lowpro30__ joined, telex joined 20:27 rurban joined 20:29 lowpro30_ left 20:30 benabik left
timotimo moarvm now has a constant pool for integer values between 0 and 15 inclusive 20:30
takes the memory usage of perl6-m -e 'say 1' down to ~89 from ~92 20:31
megabytes*
20:31 lowpro30 joined
PerlJam timotimo++ every little bit helps! :) 20:32
jnthn m: say 89/92
camelia rakudo-moar 6a6a5b: OUTPUT«0.967391␤»
20:32 jnap joined 20:34 lowpro30__ left 20:36 jnap left
FROGGS good evening 20:37
masak evenin', FROGGS 20:38
timotimo hey FROGGS
jnthn o/ FROGGS 20:39
FROGGS hey, what a greeting :o)
timotimo i kind of feel like i've drained my energy resources for the day 20:40
feel free to entertain me with cool commits, though :) 20:41
FROGGS hmmm, I will probably only send a PR for File::Find to make its tests pass under windows
not very enternainy :/
dalek o: 25c12a6 | masak++ | bin/ufo:
[bin/ufo] fix typo

  itz++		 20:42
20:48 benabik joined
FROGGS huh, perl6-m builds faster on windows in a vm than on my linux host O.o 20:49
jnthn o.O 20:51
FROGGS it only took 47s for stage parse and on my linux it usually takes 51s 20:59
21:03 Rotwang joined
lue FROGGS: what's the margin of error for those times? :) 21:03
FROGGS the what? I dont event understand the question :o)
lue en.wikipedia.org/wiki/Margin_of_error 21:05
jnthn FROGGS: Did you applied most holy statistical tests to your results? :P
lue (don't know if that would help you understand :P)
FROGGS ummm 21:06
FROGGS continues smiling :o)
lue FROGGS: basically, is the difference between 47s and 51s significant, or could that just be within the realm of fluctuating times.
FROGGS lue: I've perl6-m loads of times on my linux box and the fastest ever was 49s 21:07
but I need to rebuild it on both machines soon, so I will have an eye on that 21:08
lue If the slowest VM time is greater than 49s, you'll know the difference isn't too significant :)
FROGGS the win7 in the vm took 47s where its host takes >49s (without a vm running) 21:15
dalek kudo/nom: ca35ec8 | (Tobias Leich)++ | src/core/IO.pm:
cleanup paths to get os specific delimeter
 21:19
21:21 rurban left 21:27 jnap joined, benabik left 21:33 [Sno] left 21:34 [Sno] joined 21:35 arnsholt left
FROGGS tadzik: I PR'd you 21:42
21:43 lowpro30_ joined 21:44 rindolf left 21:47 lowpro30 left 21:53 lowpro30__ joined 21:54 sisar left 21:57 lowpro30_ left 22:05 Alina-malina left, Alina-malina joined 22:06 pecastro joined 22:09 immortal joined, immortal left, immortal joined 22:10 kbaker left, erkan left 22:13 skids left 22:14 Zaydon left
tadzik hmm 22:15
ok, alright :) 22:16
FROGGS :o9
tadzik I tagged panda as 2014.01, and will update the submodule in the newer panda
FROGGS :o) *
cool! 22:17
22:17 denisboyun joined
tadzik so it still works for people using released rakudo 22:17
22:17 zakharyas left
dalek nda: 934a4f3 | tadzik++ | ext/File__Find:
Update File::Find
 22:18
22:19 lowpro30__ left
tadzik FROGGS++ 22:20
FROGGS now there might only be the STable problem left to fix... 22:21
22:21 grondilu left 22:24 virtualsue joined
jnthn FROGGS: I'm busy on IO things, but I can give you a hint on how to gather some info on what the problem might be there, if you like? 22:24
FROGGS sure I do!
it does not seem to be about augmenting or custom infixes this time btw 22:25
jnthn FROGGS: gist.github.com/jnthn/a09d2b3beb70866be1bd
Add that patch
You can't build Rakudo itself with it applied (it legit triggers this)
But you should be able to try going through the Panda build 22:26
When it blows up, use --ll-exception to get a backtrac
FROGGS k
jnthn *backtrace
22:27 colomon joined
colomon o/ 22:27
vendethiel Is there a binary unpacking syntax in perl ? Could that be done with macros ?
FROGGS vendethiel: Buf has a method unpack 22:29
raydiak hi colomon
jnthn o/ colomon 22:30
colomon any excitement in the p6 world today? I've been awa from the Internets
jnthn timotimo++ came up with an optimization in Moar that shaved 3MB off Rakudo's base memory 22:32
And likely has performance benefits too
22:34 grondilu joined
FROGGS jnthn: it might be related to an enum State <...> 22:37
jnthn Hm 22:38
FROGGS jnthn: I will do a lot of testing tomorrow with this
jnthn Got a backtrace?
FROGGS yeah
jnthn Somewhere in EnumHOW?
FROGGS gist.github.com/FROGGS/ba492ddbaa876aaf9aed
more in Perl6/Metamodel.moarvm:add_method 22:39
jnthn um, huh
That's not the exception we added though...
FROGGS true
22:41 virtualsue left
FROGGS perhaps our added exception got swallowed by a try/CATCH or it happens in a shell("$*EXECUTABLE_NAME[...] 22:41
I will play with it tomorrow
and then I will report to you :o)
jnthn ok :) 22:42
FROGGS++
22:43 denisboyun left 22:45 denisboyun joined 22:55 rurban joined
raydiak knows better than to ask for opinions on his designs (especially incomplete ones), but must be in a reckless mood today... 22:58
p6doc.cyberuniverses.com/ # is this pleasing or ugly?
PerlJam It's not ugly.
It looks nice.
Not as colorful as camelia :) 22:59
raydiak couldn't decide if it's soothingly soft, or bland in that "lukewarm water, spit it out" kinda way :)
wasn't sure how colorful/contrasty to go, since it's documentation (local html docs will look like this too, for now) 23:00
23:00 dmol left
lue raydiak: could/should be more colorful (esp. if it goes on the perl6.org domain), but the layout looks great. 23:01
jnthn raydiak: Thanks for looking at making doc.perl6.org more attractive. The search facility stands out a lot better wiht this.
raydiak jnthn: aye, that was one of the goals 23:02
dalek rl6-roast-data: 9fce7cb | coke++ | / (3 files):
today (automated commit)
 23:03
jnthn raydiak: One thing I note is that the link/heading color scheme is blue, but the rest is green. I wonder (but I'm *not* a designer!) if tweaking the links/heading colors might be nice also.
It doesn't look that bad with them different, though.
raydiak yeah I thought about that...wasn't sure either 23:04
jnthn I'd maybe make the Search word / box a little larger. Or maybe just wider.
raydiak PerlJam, lue, jnthn: thanks for the input, I'll play with the colors some more
I could make it larger, but I'd also like to avoid wrapping...not really happy with how the header looks when it wraps yet 23:05
jnthn What I think might be nice in the end is if we could have a left hand "navigation" area for browsing.
So if you're looking at a method in a certain class you see other methods in that class to browse to
PerlJam aye
jnthn That's a much bigger thing, though. 23:06
lue raydiak: just steal the pastel color values from perl6.org, no problem :)
raydiak ah, that would be nice
PerlJam raydiak++ 23:07
raydiak lue: that *would* be a sensible approach, wouldn't it? no wonder it didn't occur to me :)
raydiak starts ripping styles from perl6.org 23:08
(or maybe the whole stylesheet) 23:09
lue IIRC I once tried (and failed) to work on a doc.perl6.org style, and I believe I just modified the base perl6.org stylesheet. 23:10
raydiak it's looking like there are at least several things I can cannibalize from it...also need to merge in parts of the perlcabal.org/syn stylesheet, which is just being linked to directly right now (I'm assuming b/c both use Pod2HTML or something) 23:13
lue yeah, though I'd appreciate a pretter perlcabal.org/syn to come out of this :)
*prettier
raydiak good idea, would be smart to put all the Pod stuff in its own stylesheet, and perhaps include it with Pod2HTML as a default or something 23:15
lue I don't think the stylesheet is linked to the generation process, but I could be wrong. 23:16
23:17 denisboyun left
raydiak Pod::To::HTML generates an inline stylesheet, AND links to the perlcabal.org one, whether you want it to or not...hacking on that is probably going to come up real soon 23:18
so far I've just been overriding it all with higher-precedence CSS rules
lue raydiak: just so you know, I have a project that I should start working on again real soon (perhaps even right now) to replace our crusty old Pod2HTML generator. 23:19
raydiak lue: I request a way to generate the content without the surrounding head/body/html tags and such...for embedding into other stuff...and a way to choose whether it spits out html or xhtml 23:21
lue I could probably flip a switch between Pod::To::HTML and ::To::XHTML, but I'm not in charge of how the generator(s) themselves work. 23:22
raydiak I didn't know the old one was old and crusty...now I know why it feels crusty :) 23:23
lue The only one who knows anything about the current one is moritz, and that's simply because he's the one who's been hammering it into working condition every once in a while :)
lue feels like working on his generator again :) 23:24
raydiak I've been reading it over, it's actually not that bad, except it's meant to only generate whole independent pages...making it hard to integrate into other non-Pod-generated things 23:25
lue My new one, once it works, will require all our current synopses to be converted to pod6 format, that'll be the best part :) [but as moritz requested, I'm not doing that until we have a generator that works]
raydiak I will hold off directly hacking Pod::To::HTML for a while and see where lue++ goes :) 23:28
have more than enough on my plate for now anyway...Math::ThreeD's matrix support is still broken, for need of like an hour of my time...and Pray hasn't been touched in like a month, which is sad 23:29
lue raydiak: again, I'm not touching Pod::To::HTML, so go ahead :) 23:30
23:30 treehug88 left
raydiak oh heck, realizing I never even posted an actual release anouncement on the Pray site :P 23:30
lue: is your generator not something which would replace it?
lue raydiak: no, I'm replacing the thing in perl6/mu that pushes all the synopses through the generation process. 23:31
raydiak ah, now I see
lue If you could make it so I could use Pod::To::HTML from within perl6, instead of resorting to shell() and the like... :P
23:32 Araq joined
raydiak lue: that's how it's being done in perl6/doc/htmlify.pl, iirc 23:32
"that" as in "being used as a module instead of a shell script"
23:33 tgt left
lue (As an aside, trying perl6-m on my generator, just to get familiar with the generator again, caused it to be Killed by kernel after taking 80% of my memory. I'll complain more once I make the code public.) 23:33
23:33 tgt joined
raydiak tried to gen the docs w/perl6-m, but is not nearly patient enough 23:34
it works, just (very) slowly
lue raydiak: ah, I guess I just didn't see what function in ::HTML I needed to call, I'll look at htmlify.pl more closely in a bit to see how to do it specifically :) 23:36
23:36 btyler left 23:37 tgt left
raydiak lue: it's pod2html, but looks like it's building up a Pod object to pass, not just source text, so it may not be as simple as we'd hope 23:39
lue If it's building up an object to send through the rest of ::HTML, maybe it could be backported to that module? Seems like it'd be useful, I know of two projects that do or would use it :D 23:40
23:42 Araq left
raydiak I'm hoping it's just a choice, and that there are facilities to do that already...if not, then yes, they certainly should exist 23:43
lue raydiak: it'd be nice if something like that were standardized, but I'm not too familiar with DOC stuff (which is where I'm guessing it would be) 23:44
raydiak well, maybe there isn't yet: github.com/perl6/Pod-To-HTML/blob/...TML.pm#L17
lue I think that comment's more about the internal programming of that module, and less so using the module. 23:47
raydiak *blink*...*reread*...yeah, I suppose I should slow down and carefully read words one at a time :)
alright, break time, and then colorfullness, and then we'll see what comes next; thanks lue++ for giving me a few future targets to start aiming at while I'm moving in those directions anyway 23:50
23:50 Adriaaan left
lue no problem :) 23:51
23:54 BenGoldberg joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!