pugs.blogs.com | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | <stevan> Moose... it's the new Camel ":P | .pmc == PPI source filters!
Set by Alias_ on 16 March 2006.
00:17 jisom joined 00:21 fglock joined 00:22 jisom_ joined 00:36 premshree_ joined 00:39 KingDiamond joined 00:50 jisom_ is now known as jisom 01:01 Nouk joined 01:05 mako132_ joined 01:10 dduncan joined
svnbot6 r9660 | fglock++ | pX/.../PCR - lexical subrules implemented, but PadWalker returns 'undef' (Windows bug?) 01:17
01:24 b_jonas joined
fglock PCR lexical subrules work fine in feather, but fail in Windows 01:25
svnbot6 r9661 | fglock++ | pX/.../PCR - cleanup debugging info 01:38
01:38 justatheory joined 01:47 justatheory joined 01:48 GabrielVieira joined
svnbot6 r9662 | fglock++ | pX/.../PCR - shorter TODO 01:50
GabrielVieira hello there
fglock oi Gabriel 01:51
GabrielVieira td bem flavio?
ops
GabrielVieira turned on english input :P
fglock trabalhando a luz de lanterna aqui, lapto na bateria... 01:52
laptop
GabrielVieira haha
serio?
faltou energia ou foi a conta? :P
fglock yes (blackout)
GabrielVieira humm
foda em
cascavel?
fglock it was raining - no, Porto Alegre 01:53
GabrielVieira hum
all the city? 01:54
or just your neighborhood?
fglock I see no lights in 1km around 01:55
GabrielVieira i was watching "Falcon, Sons of the trafic" on globo :D
humm
fglock i was watching lord of the rings ...
GabrielVieira i dont like the translated version ;~ 01:56
just with legends or nothing else
fglock look at this noticias.uol.com.br/ultnot/efe/2006...26877.jhtm 01:59
fglock very OT, isn't it? 02:01
GabrielVieira yep 02:02
i didnt know about that
:~
fglock are you going to YAPC::Brasil? 02:03
GabrielVieira not this year :/ 02:04
maybe next one :D
02:06 Lemgar joined 02:13 mako132_ joined, revdiablo joined, jiing_ joined, jiing joined, mugwump joined, SamB joined, spo0nman joined, knewt joined, nnunley joined, Khisanth joined, PerlJam joined, avar joined, samhain1138 joined, p5evalbot joined, buu joined, Aragone joined, lisppaste3 joined, hlen joined, Lorn__ joined, arcady joined, LeTo joined, Juerd joined, audreyt joined, azuroth joined, theorbtwo joined, rafl joined, Southen joined, Eimi joined, kane__ joined, hcchien joined, tewk joined, jdv79 joined, scw joined, qwacky joined, pasteling joined, rep joined, s7n joined, Odin- joined, Steve_p joined 02:14 fglock left 02:38 xinming joined 02:39 xinming joined
xinming hi all, I'm in Kunming now. :-) 02:43
merlyn where is kunming? 03:13
ayrnieu worldfacts.us/China-Kunming.htm 03:21
homepage.mac.com/johngoodman/
www.maps-of-china.com/kunming-ow.shtml 03:23
merlyn Oh. somewhere in china
someplace I'm not likely to go in the next ten years
unless they decide they want the US. :)
ayrnieu (better: www.maps-of-china.com/china-country.shtml)
merlyn ... en.wikipedia.org/wiki/Kunming 03:24
wikipedia++
ayrnieu it needs a better map, though.
well, you can squint at one of the maps on /China 03:25
merlyn still - not likely I'll end up there. 03:29
as bad as the US is getting for human rights, at least we aren't China. :) 03:30
ayrnieu it's probably safer to be !chinese in China.
merlyn Even a felon?
ayrnieu what? You're not a sex offender. 03:31
merlyn True.
I'm offensive during sex, sometimes. :)
but that's not the same thing at all.
xinming Hmm, the best introduction for kunming is there is no other seasons except spring. :-) 03:47
merlyn too bad it's still "the land of the opposite of free" 03:49
I am sad for 2 billion people 03:50
but thankfully, this next rum and coke will help me forget that
04:04 KingDiamond joined 04:05 arusa joined
arusa human rights? # www.thenation.com/doc/20051226/klein 04:06
oooops.....wrong channel ;)
04:07 arusa left
ayrnieu actually, right channel. 04:15
making assumptions of that URL.
04:34 masak joined 04:54 amnesiac joined 04:56 kisu joined 05:08 elmex_ joined 05:13 eric256 joined
eric256 hey....a bit OT but....lol...does anyone here know of a good photo tagging package? 05:15
it seems like there would be hundreds out there since digital photography is soo huge, and yet, i don't realy see much out there.
mugwump eric256: f-spot 05:16
has a nice gui, and it's in ubuntu
eric256 yea i saw f-spot but it is gnome only...should have mentioned i'm on windows ;)
mugwump um.. flickr.com ? :) 05:17
eric256 hehe yea if i want to upload my 20k pictures lol. realy i should probably delete 19k but they are soo cheap to store...now if i could only find them agian
i was thinking though...if windows had a way to make links...a software could store all the pictures in one spot, then build fake directories where each directory was a key word and put links to all the pictures in each directory that matched there tags. then just have it scan for real files being added, so i just put files into the right directories, and it pulls them into the archive and puts links in all the right spots. hmm 05:19
ayrnieu If you have 20k pictures, you'll need a mechanical turk just to tag them. 05:23
eric256 aint that the truth
eric256 starts windows counting his pictures out of curiosity 05:24
ahh a disappointing 6.7k oh well. still quite a few.
diotalevi Ah, eric256 reinvents the DB as a file system. :-) 05:26
ayrnieu I recommend that you get to work on the turk, then, and process it in off moments -- or think of a way to distribute the processing (mainly, you want to protect yourself from chaotic tagging).
eric256 diotalevi if windows would come out with there filesystem that is suppose to handle this then life would be good. ;
diotalevi eh. that's supposed to be vista. I've wanted this for linux for years. 05:31
eric256 should be easy on linux
as far as i can tell the scheme i just cooked up wouldn't even require a gui . for instance you save in \vacation\2005\ and in also shows up in \2005\vacation ;) and even in just \2005 or \vacation. save in any directory and it automagicaly tags with all the directories as keywords. hardest part is probably watching all the directories to see new additions and modify links/directories as needed 05:32
diotalevi say, you know windows does symlinks, right? 05:36
ayrnieu ... if you're already encoding the tags as directories, why watch directories and automatically tag things?
eric256 i've been reading on it yes. and it sort of does. ;) shortcuts arn't symlinks, but there are some utilities that do, but only for directories 05:37
ayrnieu because sometimes i'll be looking in 2005 and sometimes in vacations, and sometimes i want all pictures with both keywords
ayrnieu eric - do you already have the images in the directories in this manner?
eric256 so the picture would need to show up in all those directories ;) unless i figure out how you make your own folders like My network places and then I could code my own system 05:38
ayrnieu....maybe 50% ;) but not as detailed as i would like, becuase then it becomes hard to find a picture unless you know *all* of its tags
diotalevi I think the Windows-ish way to do what you want is with one of those "this diretory works like a web page" things and you do some DHTML. 05:39
ayrnieu eric - then you've already a significant headstart on a real solution. A real solution might also make use of a local webserver.
FurnaceBoy_ (hearts) Flickr
ayrnieu FurnaceBoy - eric256 doubts that it can handle his 6.5K images. 05:40
eric256 i've thought about both of those, but i want any program looking at the pictures DWIM and not see the underlying structure.
so my base picture folder would have one subfolder for every tag and all photos, then you narrow it down by choosing sub categories (and maybe pictures arn't visible till you choose at least one tag) 05:41
hehe windows has fsutil for creating hardlinks and hlscan to scan them...now to start work ; 05:47
)
spo0nman eric256: you can write a script in 10 minutes. 05:50
eric256: for tagging and uploading to flickr.
eric256 spo0nman and how does that help me access them on my own computer? 05:51
flickr is nice and all, but i'm not real comfortable giving them all of my pictures to store and tag ;)
spo0nman eric256: hmm ...
eric256:I work for them ... we are not evl. 05:52
:)
eric256 okay...still doesn't solve my problem lol
now if you have a local version coming out that convinces the normal windows FS that it is realy a tagged media album i'll be happy to use it 05:53
spo0nman eric256:yeah! well you'd have to write a db thing in sqllite then.
eric256:you're on windows?
eric256 for this project ;)
i wonder if its easier to watch the directories for changes....or just scan periodicaly 05:56
ayrnieu depends on the platform. 05:57
eric256 win xp ;)
05:58 arcady joined
spo0nman i would do a database with a simple schema photo->location photosids->tag-ids tag->ids->tagname. 05:59
eric256 spo0nman that is not a hard part ;) the only think i'm working on is makeing a regular windows folder system pretend that it is actualy a media tagging system 06:00
06:00 azuroth joined, Southen joined
eric256 Win32::ChangeNotify will let me do the job. ;) now to figure out how to get it installed. lol. oddly ppm doesn't find it. probably screwed up my repos agian somehow 06:01
spo0nman eric256: ah!
06:13 mj41_ joined
eric256 now i've got all the peices and its time to sleep...bahh. liked it better when i could pull all nighters ;) wife would probably smack me if she woke up at 5am an i was still up. ;) later all 06:14
06:20 eric256 left 06:22 Southen_ joined 06:23 particle_ joined
spo0nman KingDiamond: Pfffff. 06:29
KingDiamond audreyt: there? 06:31
audreyt KingDiamond: I'm still massively jetlagged 06:35
expect me to back to normal mental capacity in... I don't know, 24 hours or so :) 06:36
KingDiamond audreyt: okay, np :-) 06:48
06:53 GeJ joined
ayrnieu how can I perform zero-width negative lookahead/lookbehinds in rules? I guessed <-after ...> <-behind ...> , but this doen't parse... 07:00
ah, ! 07:01
also, that should've been <!before ...> 07:02
svnbot6 r9663 | Darren_Duncan++ | r3235@darren-duncans-power-mac-g4: darrenduncan | 2006-03-19 23:02:28 -0800 07:04
r9663 | Darren_Duncan++ | ext/Rosetta/ : continued rewrite of Language.pod
07:12 Toaster joined 07:14 Soga joined 07:16 iblechbot joined
ayrnieu makes pretty LJ-url-matching rules from his perl5, only to discover in the process that the negative lookahead was perfectly useless (but harmless) all along. 07:16
07:38 drrho joined 07:39 Quell joined 07:51 marmic joined 07:57 pdcawley joined 08:00 xern joined 08:01 nothingmuch joined 08:43 ghenry joined 08:51 xinming joined
GeJ morning folks 08:55
08:56 rgs joined 09:05 Toaster joined
ingy hola 09:05
dduncan yello
Supaplex nite 09:06
svnbot6 r9664 | Darren_Duncan++ | r3239@darren-duncans-power-mac-g4: darrenduncan | 2006-03-20 01:06:22 -0800 09:07
r9664 | Darren_Duncan++ | ext/Rosetta/ : continued rewrite of Language.pod
ingy good night Supaplex 09:08
hola dduncan
dduncan ditto
nothingmuch møøsage
dduncan feel free to wade into Language.pod, though what I've written so far is the 80% of the 10,000 mile view so everything may be too abstract to be useful ... still, lots of revolutions in there 09:10
ingy nothingmuch: want to help me figure something out with Module::Compile 09:13
using SEE
09:23 Aragone is now known as Arathorn
Arathorn morning everyone & everything 09:27
nothingmuch ingy: not now... 09:28
i'm at work
and stressed ou
t
KingDiamond nothingmuch: looks like I'm gonna be out of luck for sometime at least then 09:36
nothingmuch home come? 09:38
it's only till wednesday
and it doesn't include the afternoon
and you didn't seem to need any help for the past few days =) 09:39
KingDiamond nothingmuch: I was not in town, and had spotty internet access :-(
nothingmuch oh
err
well, others can also help
and I can be pinged for 5 minute increments
KingDiamond yeah, I guess :)
heh 09:40
nothingmuch just not for deep magic hacking sessions like ingy wants =)
KingDiamond heh
nothingmuch to help with M::C i need to do a full context switch and *really* think about something else
ingy back 09:41
nothingmuch: so... ready? 09:42
09:44 dduncan left
nothingmuch ingy: no... big deadline 09:44
i'll make it
but I want to be 100% sure
maybe later today
ingy someday when /me and nothingmuch are flatmates, pairing will be easier 09:45
nothingmuch =) 09:46
do you plan on coming back from .tw to .us?
heh. to .com
=)
ingy nothingmuch: at times, yes 09:47
nothingmuch where do you intend on living?
ingy .cai
nothingmuch cai? 09:48
ingy The Sovereign Republic of Coffee And Internet
09:49 Samhain joined
nothingmuch hehe 09:49
in that case if you let me know in advance and we can arrange i don't see why we can't be flatmates 09:50
i mean, it's not like you're a weird rapist or something
you're just weird
ingy Samhain!
nothingmuch .... has quit
ingy we are 138
nothingmuch 138?
ingy google: Samhain 138 09:51
nothingmuch oh
09:51 Samhain joined
ingy www.onethirtyeight.com/main.html 09:51
I think Samhain predates nothingmuch 09:52
nothingmuch yes, by a lot
ingy I bet Samhain probably predates Samhain
nothingmuch but most the music I hear predates me
rgs isn't samhain the french name for halloween, btw 09:53
ingy aye
sah-wane
nothingmuch rgs: i thought you were supposed to know ;-)
ingy or somesuch
s/somesuch/somesuchshit/g 09:54
09:55 Toaster joined 10:00 KingDiamond left, KingDiamond joined 10:33 kane_ joined 10:35 Toaster joined 10:49 mj41_ joined
svnbot6 r9665 | fglock++ | pX/.../PCR - fast grammar 10:58
11:07 mauke joined 11:10 nothingmuch joined 11:14 kanru joined
spo0nman Im trying to install pugs on Freebsd 11:20
and I've got parrot-0.4.2 ghc 6.4 and perl 5.6.1 11:21
I am facing problems while compiling pugs.
this is the output of my steps after I downloaded hs-plugins. 11:22
pastebin.com/612147
it'll be nice if someone explains what is it that im doing wrong.
nothingmuch you don't need hs-plugins unless you want 'eval_haskell'
just FYI
spo0nman nothingmuch: hmm where do i get that?
nothingmuch: can you look at the errors please?
nothingmuch looking 11:23
i think you ran setup and friends wrong
but I don't know how to do that right
spo0nman pugs errors are right down at the end.
nothingmuch ask on #haskell
unless someone here steps up
mauke hmm, missing symbols from gmp
nothingmuch but in general if you're just poking around, start without hs-plugins for now
spo0nman how to start without them? i already installed it. 11:25
11:42 renato joined 11:55 foosroo joined 12:14 siosiosios joined 12:16 chris2 joined 12:31 Lorn_ joined 12:34 iblechbot joined 12:53 Khisanth joined 13:05 fglock joined
svnbot6 r9666 | fglock++ | pX/.../PCR - built-in rules: alpha, alnum, ... 13:08
Juerd seen audreyt 13:17
No seenbot? :|
13:18 jisom joined
nothingmuch Juerd: no =( 13:21
svnbot6 r9667 | fglock++ | pX/.../PCR - cleared warnings in t/04-rule.t
13:21 jp-autark joined
fglock does anyone have experience with PadWalker? 13:27
broquaint A little. 13:28
fglock broquaint: I can find a variable, but the value comes 'undef' - I'm using peek_my() 13:29
broquaint A code snippet would be handy, fglock :) 13:30
fglock do you have a pugs tree? /misc/pX/Common/Pugs-Compiler-Rule/lib/Pugs/Runtime/Rule.pm - last subroutine in this file. I'll paste it here, one sec. 13:31
broquaint I've got it, I'll take a look now. 13:32
pasteling "fglock" at 200.17.89.80 pasted "non-working peek_my()" (33 lines, 498B) at sial.org/pbot/16412
broquaint And what are you trying to achieve? 13:33
fglock you can test it by running t/04-rule.t - I'm trying to get the value of a lexical variable in the user's program 13:34
broquaint Right-o.
fglock uncomment the warn() in line 343, and it'll show the variable contents (I get undef) 13:35
broquaint Hrm ... 13:40
My subconscious recognises the problem but it's not quite making it to the conscious level yet ... 13:43
Declaring $match twice can't be helping, fglock ;) 13:44
Can't call method "match" on unblessed reference at (eval 36) line 7. 13:46
Which is why $match is empty because an error is occurring on a previous line.
13:47 pmichaud joined
pasteling "broquaint" at 217.206.131.214 pasted "Broken perl5 code in generated rule (the match called fails as the method can't be found)" (2 lines, 273B) at sial.org/pbot/16413 13:49
13:50 bsb joined
broquaint So the perl5 line in that pasted code is a kv in the rule that is generated. The error occurs there. I'm guessing the peek_my bit is working though. 13:50
fglock broquaint: $match fixed, thanks - but I still get undef - $r is what peek_my should be looking for - and it returns undef :( 13:51
broquaint Are you sure it's undef? 13:52
Check $@.
pasteling "fglock" at 200.17.89.80 pasted "debugging get_variable()" (38 lines, 648B) at sial.org/pbot/16414 13:54
fglock $@ is clear in get_variable() 13:55
broquaint Yeah, I'm having problems debugging :/ 13:56
13:59 Qiang joined
broquaint It looks like it's returning something. 14:05
Got it working now. 14:06
fglock can't wait :) 14:07
pasteling "broquaint" at 217.206.131.214 pasted "The working get_variable()" (12 lines, 239B) at sial.org/pbot/16415 14:09
fglock broquaint: it doesn't work here 14:12
broquaint Ok, I'll commit my working stuff shortly. 14:14
fglock broquaint: which version of PadWalker are you using? 14:15
broquaint 1.0 14:18
I needed it for Params::Named.
fglock 0.10 here - I'll try upgrading 14:20
broquaint Good call :) I've commited the fixes now too.
svnbot6 r9668 | broquaint++ | * Fixed lexical variable lookups. 14:22
fglock broquaint: thanks - it looks like this module will require 1.0 14:23
broquaint Ah well, at least the variable lookup should be more robust now. 14:24
svnbot6 r9669 | broquaint++ | * Fixed bug where a lexical scope might be empty and the lookup would end.
14:25 vel joined
spo0nman merlyn: foo. 14:29
14:33 diotalevi joined 14:43 elmex joined
svnbot6 r9670 | fglock++ | pX/.../PCR/TODO - make it depend on PadWalker 1.0 14:46
14:46 taeli joined 15:19 FurnaceBoy_ joined, hexmode joined 15:36 hexmode` joined 15:47 hexmode joined
stevan Juerd: ping 15:49
Juerd pong
stevan Juerd: I need a recomendation for a good ergonomic keyboard for my wife 15:50
I know you are familiar with such stuff :) 15:51
15:56 amnesiac joined
obra heh. 16:02
16:06 justatheory joined 16:10 taeli- joined 16:12 Qiang joined, fglock joined, Lorn_ joined, renato joined, KingDiamond joined, masak joined, stennie joined, ycheng joined, cognominal joined, merlyn joined, TimToady joined, dvtoo joined, szbalint joined, weinig joined, f0rth joined, beppu joined, gaal joined, ajs_home joined, pjcj joined, lumi joined, cookys joined 16:30 FurnaceBoy_ is now known as FurnaceBoy 16:39 vborja joined
Juerd stevan: Depends on what the problem is, but in general, I'm very fond of my Kinesis keyboard 16:48
stevan Juerd: she is mostly have wrist discomfort, and numbness in the fingers
Juerd stevan: And sorry for my abrupt and unannounced leaving; had to assist someone who cut off part of his index finger, and bring the guy to hospital 16:49
obra stevan: what's she using right now?
stevan Juerd: no problem :)
obra (Has she seen a doctor?)
stevan obra: the default keyboard which came with the iMac
obra: no, but she is going to soon
obra good. that's step one
xerox What iMac, if I may ask?
Juerd stevan: Then a Kinesis will probably help. Have her see a chiropractor or kinesiologist (speling guessed)
stevan she used to use a laptop,.. which is when the problem began 16:50
Juerd stevan: ... especially if she also suffers from cold hands (bad blood circulation)
stevan xerox: the lastest G5 model
Juerd: cool,.. thanks :)
obra even a $30 MS Natural is better than the default thing. Kinesis keyboards are also good for lots of folks. If she's really far gone, a DataHand may be interesting
Juerd stevan: Also, make sure she learns dvorak... :)
stevan Juerd: yes, I was going to suggest that to her 16:51
obra and if it looks like she needs a datahand, you're welcome to borrow mine to try out before you consider a $800 keyboard
stevan obra: yeah, she was thinking about the natural one
Juerd One thing about Kinesis... You *must* clip your nails more often than with other keyboards, and it's not usable by people who like to have long fingernails :)
stevan datahand == funky :)
obra finds that a ms natural works better for him than a kinesis 16:52
Juerd obra: I've often looked at the datahand online; it's indeed extremely expensive. Would you say it's worth it?
obra also, how's her typing posture? how close is the keyboard to her lap?
Juerd: for a couple folks I know, the choice was "datahand or find a new career"
and it saved their careers.
For me, it was worth it, only because my employer paid for it 16:53
the ms natural elite is still my all-time favorite keyboard.
moreso, even than an IBM clicky ;)
Juerd MS Natural and Kinesis differ in one important aspect: A "natural" keyboard forces slim people to widen their elbows, and that eventually starts hurting your shoulders.
And Kinesis keyboards are too narrow for fatter people
obra: I see... 16:54
stevan obra: her posture was horrible (when she was using the laptop), but is better now that she is on the desktop
obra carries a ms natural with his laptop
Juerd Laptops are great only for weak hands. Everything else starts to hurt then :)
And, obviously, most laptop keyboards suck incredibly (all the cheap brands, at least) 16:55
stevan is a light typer, so has never had an issue
obra for me, it's about wrist position
stevan but when i paint/draw, i tend to hold the brush/pencil very tight and get numbness easy
Juerd obra: The natural position for wrists depends on how far apart your shoulders are..
For me, the Kinesis gives me a more natural wrist position; I've used several "natural" keyboards before I figured that out. 16:56
Including MS Natural Elite
Three different Logitech ones, two MS ones, one white label one. 16:57
obra has wide shoulders
Juerd I liked the original MS Natural keyboard best.
obra Also, its biggest problem: the kinesis sucks to carry around and/or use in my lap
Juerd stevan: Look for ergonomical brush holders. They're silicone grips.
stevan Juerd: interesting,.. I will 16:58
Juerd stevan: The painter who painted our house used those
That's of course very different from painting art, but my guess is that such a solution probably works for both kinds.
obra: Agreed. It absolutely sucks for non-stationary. 16:59
I mostly work in one place, though, and the little time I use my laptop, I'm able to use my (splendid) IBM laptop keyboard.
obra I also have an IBM keyboard that's "the top bit of a thinkpad"
Juerd In fact, I like it very much because it has 96% keys compared to desktops and other laptops. That's great for my small hands.
obra but again with the "wrong angles" 17:00
Juerd I wish there was an IBM UltraNav *split* keyboard.
obra but it has a trackpoint, touchpad and thinkpad keys. and a USB port
to my knowledge, IBM has only done one split keyboard
Juerd obra: That's the UltraNav.
obra and 20 years later, they go used for > $300
Juerd The Kinesis would be perfect (except for carrying) if it came in two parts.
obra and had a trackpoint 17:01
Juerd The fixed distances and angles are SO stupid.
obra Developing for the web without a mouse is a bit of a pain
But I need to run off and do this annoying "work"
Juerd Have "fun" 17:05
17:06 stennie_ joined
obra "Thanks" 17:07
17:08 taeli- is now known as taeli 17:10 Samhain is now known as samhain1138
samhain1138 rgs: samhain is the pagan holiday from which halloween is derived 17:11
rgs pagan ? you mean celtic
17:13 taeli- joined 17:17 taeli- joined
samhain1138 ingy: Samhain were still playing after nothingmuch was born ;) 17:18
ingy yep
samhain1138 ingy: what's really sad is that i never got to see the misfits, and i'm a huge fiend (tattoos and all ;))
17:20 DesreveR joined
ingy samhain1138: I'm sure there are plenty in that boat 17:20
17:32 ruoso joined
ruoso hi all 17:32
fglock, news?
fglock ruoso: hi!
ruoso: PCR is much faster! I fixed the unneeded backtrackings 17:33
ruoso fglock, nice...
fglock ruoso: and it passes all OO, subrule tests
ruoso fglock, I'm not at my computer now... is it ready to lrep-compiler?
fglock, nice 17:34
if it's all ok, we can just change the emitter to use the new engine
17:34 DesreveR is now known as r0nny
ruoso and bootstrap it again 17:34
then we can throw away the old code
right?
17:35 SamB joined
fglock ruoso: you may need to rewrite the return blocks, using match objects - see the PCR own grammar 17:35
ruoso: PCR is running 100 tests/second here 17:36
ruoso there are only two or three return blocks that manipulates the match object
fglock, most of them just return { lalala => $() }
fglock ok - easier then :)
gaal wow, Test::Pod::Coverage. what an excellent example of a good idea gone silly.
ruoso fglock, so... do you think PCR is able to compile Grammar/Perl6.p6 in lrep-compiler? 17:37
osfameron gaal: I think the annoying thing is the test failure on less than 100%... that should be configurable 17:38
ruoso fglock, or should I move most of the rules to some class inside PCR
like Pugs::Grammar::Statements
fglock ruoso: PCR can't compile Grammar - it compiles only Rule
ruoso fglock, hmmm
fglock, so I won't be able to use PCR as an in-place substitute for the current lrep engine? 17:39
fglock ruoso: it replaces p6rule / p6rule_lib
ruoso: it replaces iterator_engine too
ruoso fglock, that's the ruleop:: methods 17:40
fglock, and the emit_rule code
right?
gaal osfameron: indeed. I just added this, so I can still use it when I do get around to documentation.
# plan skip_all => "Test::Pod::Coverage is a little too draconic for my taste" unless $ENV{BITE_THE_BULLET};
of course, I'm likely not to remember to do that!
which is why this is silly.
fglock ruoso: yes, plus half of the grammar file
ruoso fglock, isn't it just a matter of including one rule and one emit node to PCR?
fglock, AFAICS this is what's missing to compile the grammar 17:41
osfameron our old smoke started off insisting 25%, then gradually raised the bar
ruoso if it can emit the methods, emitting the class is easy
fglock, should I sublass PCR? 17:42
fglock ruoso: no, you are supposed to use or inherit from PCR - 'grammar' is not part of the 'rule' language
ruoso fglock, hmmm... maybe PCR and lrep should inherit from the same thing...
fglock ruoso: you'd better create a new Grammar, and call PCR to parse the '...' in 'rule {...}'
ruoso fglock, ok 17:43
fglock, now it makes more sense to me
fglock, so PCR can parse and emit the code for the <rule>
gaal osfameron: yeah. but the module provides no way to set a threshold. draconic.
fglock ruoso: yes
ruoso fglock, can I call it in two phases? parse first, emit later? 17:44
fglock ruoso: yes - see the implementation of compile()
ruoso fglock, nice... 17:46
ruoso just waiting to get to his computer to refactor lrep-compiler and use PCR 17:48
17:49 pmurias joined
ruoso pmurias, hi.. 17:49
pmurias hi 17:50
ruoso pmurias, any new idea on the grammar? 17:51
fglock pmurias: pX/PCR is already working - it's much faster now
ruoso pmurias, ready to lrep-compiler :) 17:52
fglock about grammar optimization - the new PCR grammar uses a kind of 'if-then-else' instead of backtracking (see the last 2 rules in the file) 17:54
pmurias fglock hinted that we should use %dispatch_table instead of @rules, now i'm trying to add %hash to PCR
fglock pmurias: eliminating backtracking makes it much faster, even without a dispatch table 17:55
see lib/Pugs/Grammar/Rule/Rule.pm 17:59
pmurias any one wants to start porting lrep-compiler to the PCR? or should i begin doing it? 18:00
ruoso pmurias, feel free 18:01
pmurias, see the recent backlog (just before you join)
pmurias, me and fglock were discussing exactly how to do it
pmurias, but I'm away from my computer and can't do it now 18:02
18:02 arcady joined
PerlJam It's too bad you don't have access to real continuations. 18:04
pmichaud yes, continuations make it *much* simpler 18:05
I wouldn't have wanted to do PGE w/o continuations
continuations allowed me to make PGE more of a compiler than an interpreter
fglock pmichaud: how do continuations help?
PerlJam You could fake it with some gotos and clever use of global data though I bet. It just requires a lot more work :)
pmichaud backtracking is easier across calls
18:05 Limbic_Region joined
pmichaud it's easier to backtrack subrules 18:05
fglock pmichaud: ok - I've used a state-machine to fake it 18:06
pmichaud fglock: right, it can be done with a state machine as well, but continuations and coroutines mean the execution trace can keep the state for you :-)
18:16 ghenry joined
TimToady pmichaud: just answered your 6p query. 18:27
pmichaud TimToady: oh, thank you thank you thank you
TimToady: yes, it's in my box
I can switch context back to parsing now instead of adding documentation and comments :-)
TimToady except that I floated a new proposal at the same time to change .() special form to _() instead... 18:28
pmichaud eek, I see that
TimToady anyway, we'll have to see what Damian says about it. 18:29
It looks more complicated, but it's actually unifying all the magic under _.
you wouldn't have to lookahead for .X things and then do different things based on what X is anymore. 18:30
pmichaud well, I was already not looking ahead for .X things -- I was switching the operator table based on the presence/absence of whitespace 18:31
TimToady gee, who told you to do it that way. :-)
pmichaud hey, it works (worked?) nicely
TimToady until you hit list operators. :) 18:32
pmichaud well, I was doing listops as a sort of prefix op, but that fell apart when the thing coming after the listop wasn't a term
TimToady You can have "foo" on one line, and the next line start with ".", and 18:33
you're still not sure whether to "eat" the space.
the current situation requires two-token lookahead, basically.
and I now think that's bad.
pmichaud oh, I was treating .(, .[, .{, etc. as tokens 18:34
TimToady Yes, but it's the same sort of token inconsistency that Perl 5 has in its sigils + whatever.
pmichaud yes
TimToady I should have recognized the design smell...
ingy TimToady: yeah, what were you tokin'? 18:35
TimToady I don't need to toke anything to be natively stupid about tokens...
ingy Maybe you had too much Tolkien 18:36
particle_ recognizes a bad pun smell
TimToady -Opungent 18:37
ingy senses a new module
-MPugent
TimToady wants to make a pun on "podule".
18:37 siosiosios joined
gaal -Mars::Poetica 18:38
particle_ podule? a plague of doc errors?
TimToady -Venus::Poetica for equal time?
gaal -Doc 18:39
ingy smells -Uranus::Poetic coming
ewwww 18:40
ingy needs to sleeep
gaal needs to finish up this module
TimToady sentences ingy to watch that segment of E.T. 100 times in a row.,
ingy E.T. skype home 18:41
pmichaud TimToady: so, if I'm understanding things correctly, essentially we handle the "terms" and "method postfix" levels of precedence using top-down notions, and the rest can be bottom-up. And we need a parameter to the bottom up parser that says "only use tokens at XYZ precedence level or higher". Sound close?
(where XYZ precedence is really "higher than list op precedence")
TimToady Yeah, that sounds about right. Only need a crude two-way partition of precedence levels, but maybe it's worth generalizing. 18:42
pmichaud at the moment it's easy for me to generalize the bottom-up parser to say "stop on a token that is lower than XYZ level" 18:43
oh, wait, it's not as easy as I thought. But I can still do it. :-)
TimToady np, just have to decide where the actual boundary is, in case people add new prec levels. I guess if they say "is looser(LIST)" then they're obviously talking about the righthand prec of list operators, not the lefthand. 18:44
(which would be infinite like a left paren.)
pmichaud right 18:45
essentially one could use is looser(infix:<< <== >>)
sorry, is looser(infix:<,>) (had it backwards)
or is equiv(infix:<< <== >>) (you get the idea) 18:46
TimToady The question is whether that makes a new list operator or something on the other side of the fence.
we could put the boundary in the interstices between list operators and the next loosest thing, I suppose.
not sure if that makes sense though. 18:47
pmichaud I think it'll all work out well enough. Okay, that tells me where to point the flashlights
TimToady should probably just shoot anyone who tries it.
flashlights are scary to E.T.s... 18:48
pmichaud the nice, elegant operator precedence parser is becoming less elegant :-|
ingy make the precedence cutoff value a global variable
TimToady icky globalses
pmichaud I was going to make it a parameter
<op_parse: infix:,> (using PGE's syntax, not p6's) 18:49
TimToady terminators to subparses should generally be parameters.
pmichaud on the other hand, if we say that method postfix and above is handled by top-down parsing, I should be able to get rid of the whitespace switch in the bottom-up parser, yes? 18:50
TimToady you could almost generalize the parameter to a smart match, which would let us match against either a set of tokens or a range of precedences (but maybe not both with the same param)
ingy wanders off to watch random Big Lebowski segments until sleep overtakes him 18:51
Arathorn would rather Hucksucker Proxy than Big Lebowski...
pmichaud TimToady: that's a very interesting approach
hmmmmm
ingy Arathorn: that's like just your opinion man... 18:52
Arathorn hehehe
TimToady yes, top-down could probably deal with whitespace issues, but maybe only if each token looks for its trailing whitespace automatically.
18:52 eric256 joined
pmichaud well, the rule that would have to deal with trailing whitespace is the term rule, so that's not difficult 18:53
TimToady or at least "knows" somehow whether there's a _ embedded in the next whitespace.
gaal we need a ))<>(( operator
TimToady don't go there...
gaal (whoever knows the reference, honk :)
pmichaud one could almost generalize the parser into a smart match, where we have a set of states and each token knows what "states" it's active in 18:54
TimToady $string ~~ grammar is already a smart match... 18:55
pmichaud yes, I was thinking at the bottom-up level
TimToady ah. 18:56
pmichaud i.e., $token ~~ state
i.e., instead having separate term and operator states, generalize to more states
I'll toy with it another time. :-)
PerlJam pmichaud: from the outside looking in, that seems to jibe well with the language you guys have been using when talking about perl6
TimToady When I was originally trying to think about how you'd emulate bottom up with a top down parser, I was thinking along those lines. But you have to construct new precedence levels as new states on the fly, and it exceeded my mental capacity... 18:57
pmichaud I think precedence level is just a parameter to the state
each token already knows its precedence level; we just need to know what level we're currently at (similar to the shift-reduce logic we have now) 18:58
TimToady yes, I realized that at the time, and shortly thereafter my eyes glazed over...
but it's definitely worth thinking about if you can, I suspect. 18:59
pmichaud I can, I'm limited only by time. :-)
TimToady stay away from margins, then...
pmichaud well, the other limitation is wanting to get something for others to look at :-) 19:00
fglock does 'rm -rf *; svn up' - wrong way here ...
pmichaud I'll spend an hour or so brainstorming it a bit; if that doesn't lead to anything I'll go to the traditional approach for now and backburner it
TimToady that's just inside-out time, pm.
pmichaud but shift-reduce parsing becomes much simpler if we can rely on top-down to handle parens 19:01
TimToady Yes, the infinities of surreal precedence are better handled top down.
eric256 shakes his head after reading the last 30min~ of log....hmmmm lol 19:02
TimToady you're supposed to nod your head, eric256... 19:03
pmurias fglock:"Can't locate v6.pm in @INC" what do I have to put in PERL5LIB? 19:04
TimToady but you've gotta realize @Larry has been talking this over for several years, and so we tend to talk in riddles.
19:04 GeJ joined
particle_ wonders idly why @Larry isn't %Larry 19:04
TimToady because we're all shifty characters.
pmichaud in fact....... (stream of consciousness).... if we say that terms and circumfixes are handled by top-down, then the only thing left for bottom-up is operators
fglock pmurias: just run Makefile.PL and it will fix
particle_ unlike much of perl, that makes sense ;) 19:05
TimToady that's how I originally thunk it would work.
pmichaud we have prefix operators that have to be recognized in expect_term, but little else
eric256 lol
well you certainly are shifty
TimToady and everyone else goes popping off in various directions... 19:06
particle_ pushes TimToady
TimToady expands his memory 19:07
particle_ fills a bucket
pmichaud wanders off to think a bit more
TimToady challenges particle_ to a wave dual.
particle_ accelerates 19:08
TimToady just so you're not an Acceloraptor.
TimToady has to wander off too. 19:09
eric256 pugs -Obad_pun
TimToady waves
particle_ nope, i'm a boson
pmurias fglock: thanks the Makefile.PL vodoo fixed it
particle_ or is that bozon?
no, wait. i'm a moron.
PerlJam a muon would be more better. 19:10
TimToady well I'm particularly wavering, but still must depart &
19:10 mauke joined
PerlJam oh, but you said you were a boson, so I guess you can't be a muon 19:11
19:13 Muable joined
gaal a mooson? 19:13
wolverian moonsoon
gaal (that was very Ob)
PerlJam What are the particles of perl6?
particle_ no, wait, i'm a string!
PerlJam not a super string? 19:14
19:14 taeli- is now known as taeli
particle_ it's the only kind i know 19:14
PerlJam Are you a 10 dimensional object with a 4 dimensional projection? 19:15
particle_ indeed 19:16
aren't you?
PerlJam I don't know, I can only observe the 3 dimensional me. 19:17
19:20 bernhard joined
PerlJam svn merge -r HEAD:{"1990-03-20"} PerlJam 19:21
Apparently I can't revert back to an earlier state.
Muable Mind if I intrude on this highly technical discussion?
PerlJam Muable: go ahead 19:22
Muable OSDC.il piqued my interest in pugs. I've downloaded and built on Windows XP (unfortunately).
Where do I start? There's an awful lot of code there.
ruoso gets back to work...
PerlJam What is it you want to do? 19:23
Muable And I don't know much Haskell (yet?)
Not sure. Something easy to begin with.
PerlJam How much perl6 do you know?
Muable Enough perl6, I think. Plenty of perl5, certainly.
FurnaceBoy spo0nman, I heard you say you work for flickr -- kudos man, it's a beautiful thing. 19:24
PerlJam something very easy would be to write tests for things that should work in perl6 but don't in pugs
Muable Speaking of tests... a lot of mine fail. Is this normal?
PerlJam Another thing you could do, would be to try contributing to lrep
Muable What's lrep? 19:25
PerlJam Muable: look in misc/pX/Common/lrep-compiler
or even misc/pX/Common/lrep
I don't think that there's an up-to-date todo list for all of the various sub-projects that pugs has spawned. 19:26
(though I wish that all of the people working on pugs related things would put todos into rt) 19:27
Muable So lrep is supposed to be the Perl6-based Perl6 parser?
19:27 elmex joined 19:28 weinig joined
PerlJam um .... s/the/a/ and really it's sort of perl5 based. 19:29
Arathorn lrep (as i understand it) is more of an experiment to see if a perl6 parser can be written really quickly in perl5, and then bootstrapped to perl6 asap.
s/perl6/itself/
Muable Interesting... 19:30
Also: Any parts of the Haskell codebase which are reasonably painless to start exploring?
PerlJam Arathorn: s/then/concurrently/
Arathorn nods 19:31
PerlJam Muable: I don't know. The haskell code has been rewritten several times since I last looked at it.
Arathorn the only problem in contributing (from my point of view, at any rate), is that it (lrep)'s currently undergoing huge refactoring to try to unify together all the different perl5 codebases involved - Pugs::Compiler::Rules (in its various repository locations), and various versions of its own implementations, all of which are all interrelated
Arathorn couldn't get his head 'round where the current state of play was, and who was refactoring what where 19:32
you almost certainly have more tuits than I, tho' :)
19:32 weinig is now known as weinig|away
PerlJam Arathorn: pugs is a perfect example of "refactor mercilessly" applied. 19:32
Arathorn sure - but it's then hard for a newbie wanting to contribute to know what to familiarise themselves with, and try to extend 19:33
if the landscape's undergoing fullblown tectonic shift the second they touch anything
PerlJam yep. The only recourse is to jump in and start swimming.
Arathorn especially if it's fragmented across multiple locations, as lrep was when i look last week
Arathorn points sadly at his waterwings
Muable Well, that's why I'm asking where's best to jump in!
eric256 Arathorn certainly is. ;) i know that feeling.
Muable Meanwhile: I get loads of the following from "make test": 19:34
*** Undeclared variable: "$count"
at D:\Perl6\site\lib\auto\pugs\perl6\lib\Test.pm line 15, column 5-13
19:34 larsen joined
Muable t/builtins/arrays/elems.t line 11, column 3-57 19:34
Test after test fails that way. Has anyone seen this before?
Arathorn not that I remember 19:36
19:36 mj41_ joined 19:43 jisom joined 19:53 Arathorn joined
fglock Arathorn: the CPAN version of Pugs::Compiler::Rules is in /perl5; the devel-version is in misc/pX; the original lrep is pX/Common/lrep; the devel version is pX/Common/lrep-compiler 19:53
Arathorn nods - i worked it out in the end 19:54
although I wasn't sure whether to steer clear until ruoso and yourself had refactored lrep-compiler to use PCR,
fglock it would be nice to have a standard place for 'devel' x 'working' versions
PerlJam everytime I see PCR I think "polymerase chain reaction" 19:55
fglock PerlJam: it kind of have the same effect on code 19:56
particle_ :)
PerlJam It chops the code up into little pieces and then swishes it around in a DNA soup of proto-code?
;)
fglock PerlJam: that's what an AST is about, isn't it? 19:57
PerlJam modulo the inherent randomness of the process, I guess. 19:58
integral with some languages you can apply the small-step reduction semantics in any order... 19:59
19:59 dduncan joined
gaal cpan-upload --user GAAL --mailto [email@hidden.address] Template-Patch-0.01.tar.gz 20:21
20:21 larsen joined
fglock keeping $/ up-to-date during a match looks expensive - I wonder how it was supposed to be implemented 20:21
TimToady As long as it looks up-to-date when you ask for it, it can be as lazy as it likes in between. 20:23
pmichaud the tricky part is backtracking over captures :-)
20:23 larsen joined
TimToady but even Perl 5 manages that. (though confusing "temp" with "let", calling both "local") 20:24
fglock mmm, return blocks are lazy already...
pmichaud: backtracking just throws data away - my problem is to build a match tree just to throw it away 20:25
but using a lazy tree might fix it 20:27
TimToady if it's any comfort, we're trying to design the Perl 6 grammar to require almost no backtracking. 20:29
so you may be doing a premature optimization.
pmichaud I basically came to the conclusion that it was better to go ahead and build and maintain the match tree for now 20:30
TimToady Perl regexen have always been somewhat "success-oriented". Rules aren't really any different...
pmichaud A basic match object in PGE just has a reference to a string, a from index, and a to index. The rest is constructed lazily as the match proceeds 20:31
(i.e., don't build hash or array components until we actually capture something there)
away again -- brainstorming operator precedences 20:32
fglock thanks! 20:33
20:35 datenreisender joined 20:41 ruoso joined
ruoso fglock, pmurias, hi... 20:42
fglock ruoso: hi! 20:43
pmurias ruoso: hi 20:47
fglock pmurias, ruoso: hi
ruoso any news? 20:48
ruoso finally got to his machine again
ruoso wonders why svn update is taking a loooooooooot of time 20:49
fglock ruoso: I'm finding out how to implement $0, $1 backreferences
ruoso: same problem with svn here
ruoso fglock, hmm... ok... 20:51
pmurias commits take eveeen longer... 20:52
ruoso fglock, rule closures are still perl5 filtered code? or they're perl 6 already?
fglock I think there is a way to make a temporary change to the match object, without saving a deep copy
ruoso: I've got an idea now - rule closures can be implemented by inheriting the rule grammar, and reimplementing the closure parser! 20:53
ruoso fglock, but closure code can be any perl 6 code 20:54
fglock ruoso: so you can define the language inside the closure
jisom make it truly multilingual?
in every way
ruoso fglock, don't you thing it's weird to have the language grammar to inherit the rules grammar?
TimToady fglock: regarding something you said earlier, %hash and <%hash> both look up literal keys by longest-token rule. The difference is only in how it treats the value. With <%hash> the hash values are assumed to be rules.
ruoso pmurias, where does Grammar/Rules.pm comes from? it looks like lrep compiled code... 20:56
TimToady so basically any macro stuffs its keyword into the key of %hash and the "is parsed" into the value, conceptually.
ruoso pmurias, but I couldn't find the p6 source
fglock ruoso: the perl6-rule-grammar inherits from rule-grammar; the p6-r-g calls the p6 compiler to compile closures
ruoso fglock, looks better :)
pmichaud fglock: there's already a :lang modifier that has been discussed to identify the language used for closures :-)
(inside of rules)
I don't think it's in S05 yet 20:57
fglock TimToady: sure
pmurias ruoso: the code to Grammar/Rules.pm is lost now..
pmichaud this is part of the reason that PGE had trouble parsing perl 6 closures -- you'd need a Perl 6 parser to handle it properly. Thus we added the {{ ... }} closure syntax
20:57 weinig|away is now known as weinig
ruoso pmurias, lost? 20:57
fglock pmichaud: ok
pmurias it consited of the rully parts of Perl6.p6, accidently overwritten it while move lrep to PCR 20:58
ruoso pmurias, before a commit 20:59
fglock pmichaud: anyway, we could desugar :lang into a Rule subclass that knows how to handle "lang" - this may make the implementation easier
pmichaud yes, this can work 21:00
fglock pmichaud: it helps break circularity, I think
ruoso pmurias, much of the rules were in Perl6.p6, right?
pmurias, they can be reverted from svn
fglock have you lost the compiler state? 21:01
dduncan is anyone finding openfoundry reeeeeeeeeeel slow?
pmurias i can recopy them if you want, but Rules.p6 is replaced in the lrep-pcr version 21:02
fglock dduncan: it's so that you think better before committing :)
dduncan I'm finding that update/pull takes several minutes to even start
ruoso pmurias, oh... lrep-pcr is new to me :)
21:03 penk joined, penk left
dduncan today, whereas yesterday it was fine ... and even now, no pings are lost, but the website is likewise slow 21:03
21:03 sandrina_m joined, penk joined
ruoso dduncan, it looks like high load 21:04
dduncan, not a network issue
pmurias ruso: it's being commited right now...
ruoso pmurias, is it a subtree (as lrep-compiler?)
dduncan I"m committing to it right now, following the eventual update end, but its a small commit
ruoso dduncan, svn uses a database, which requires much more memory and cpu than rcs files 21:05
pmurias yes, it's dosn't fully work yet so I branched it untill rule are fully working in it 21:06
s/rule/&s/ s/untill/until/
dduncan ruoso, I understand that there has been some slowness in the past, but right now it is quite different than any past experiences ...
of course, if the server has a high load, with lots of ram being used by other things, that could perhaps cause this, if svn is being starved 21:07
ruoso dduncan, that's my bet
21:08 penk joined
ruoso does someone can help svn machine? 21:10
it is really in a bad mood..
fglock writes down the $/ strategy - hoping it makes sense... 21:13
ruoso pmurias, did you commit? I couldn't find it... 21:15
21:16 siosiosios joined 21:18 hcarty joined, hcarty left
pmurias ruoso: waiting for the authentication prompt :( 21:19
ruoso welll... lrep still can bootstrap itself... I think Rules.p6 can be recovered from svn... 21:20
fglock adds one line of code to the rule engine - after many unsuccessful attempts :) 21:21
ruoso we can just change the emitter to use the new engine and bootstrap again...
fglock ruoso: there is a "bridge" emitter in PCR/lrep - I use it to compile the PCR grammar 21:22
(please don't change PCR/lrep) 21:23
ruoso fglock, lrep in lrep-compiler can just be changed to use PCR as its rule runtime
fglock, I do think this simple change + make bootstrap will make it just works...
fglock ok 21:24
ruoso but I have to go now... maybe I'll be back later...
too much work :(
chris2 aoki writes a haskell book? w00t!
ruoso bye 21:25
pmurias i'll try to commit lrep-pcr again
fglock ruoso: tchau!
ruoso pmurias, I'll take a look later
fglock, falow
chris2 ewrongchan :P 21:26
21:31 larsen_ joined
fglock if I change '$match = &func($str)' to '&sub($str,$match,$leaf_ptr)', then the matching nodes get the current $/ for free, and backtracking still works 21:36
21:37 ruz joined 21:43 hcchien joined
Arathorn gets very confused whilst trying to use his windows desktop until he realises that he killall'd explorer in order to free some ram :\ 21:43
21:46 wilx joined
pmurias good night 21:47
fglock when you start writing things like [@{$$leaf->{state}}] there must be something wrong ... 21:51
amnesiac yes, your brain most of the time :/ 21:56
fglock amnesiac: yep 21:58
21:59 penk left
TimToady dduncan: it may be that 1000 copies of Fedora Core 5 crossing the Pacific simultaneously is what you're seeing on the network. 22:00
dduncan hello 22:01
but the problem doesn't seem to be network, as ping worked fine
TimToady Especially if a bunch of USian people are downloading it from Asia on the theory that it's the middle of the night there...
integral pings can have different performance to other data unfortunately
TimToady just a guess, anyway...
does the server serve up Fedora distributions? 22:02
dduncan I will mention that performance to many other internet servers, including one in New Zealand, is normal
TimToady It could certainly be something else.
but an ftp server allowing too many connections could bog things down. 22:03
dduncan well, a bright side of using svk is that I can commit locally and merge later, though the merging sometimes has unplesant side effects 22:08
so server outages are less of an issue
fglock do subrules backtrack when the caller rule backtracks? (I expect they do) 22:09
22:09 jisom joined 22:12 larsen joined 22:13 m joined
fglock is this an error? / $1 (.) (.) / 22:16
jisom would / $10000 (.) (.) / be realistic to check against? 22:17
fglock jisom: maybe (lazily) 22:18
jisom although to support /$1 (.)/, it'd have to do some weird backtracking and be really slow
pmichaud well, in / $1 (.) (.) / the value of $1 isn't set when it's matched, yes? 22:22
ayrnieu it's normal for equivalent expressions to have different performance characteristics, I think.
jisom pmichaud: so it'd be matching against an undefined value?
pmichaud jisom: that's what I would expect
jisom s/against/with/
that's the easy and logical way out 22:23
fglock pmichaud: do rules backtrack into subrules?
pmichaud fglock: yes, they do
to avoid the backtracking, one cuts the subrule: / <subrule>: /
ayrnieu matching-against-an-undefined-value is what perl5 seems to do. 22:24
Juerd - www.pckeyboard.com/Kustomizing.html
fglock pmichaud: ok! just checking...
mauke hmm? perl5 simply interpolates variables
ayrnieu which simplistically has this effect :-)
pmichaud the perl5 equivalent of / $1 (.) (.) / would probably be / \2 (.) (.) / ... and I dunno what that does 22:25
mauke I think it unconditionally fails
at least that's what ploki does :-)
TimToady it might turn the \2 into \x02.
22:25 b_jonas joined
jisom I thought it was \02 22:25
TimToady same thing
jisom not for \10 22:26
ayrnieu it doesn't appear to, TimToady.
TimToady might have an exception for single digit \n things
don't remember offhand.
ayrnieu ( perl -le '$_ = "\x02kllo"; print "yay" if /\2(k)(l)/' )
pmichaud I think single digits were always treated as backrefs
Arathorn didn't the behaviour of that change in one of the minor revisions of perl5?
jisom I suppose there's a good reason the perl5 regex are being somewhat dropped 22:27
Arathorn (just when the entirety of the rest of the world adopts them as standard ;D)
TimToady in any event, you still have to handle the undef case for /(a(.)|b(.))\2/ and such.
ayrnieu Arathorn - better than what they were used to.
TimToady ironical, ain't it. :) 22:28
jisom yeah, using pcre for perl6....weird
mauke (see also perlmonks.org/?node_id=424865 )
ayrnieu but if /$1 (.) (.)/ is just slow, that's alright: that's an opportunity for some{one,machine} to come along and say "hey! Your regex is pointlessly slow. You should say /(.) (.) $0/"
b_jonas mauke: that reminds me to www.perlmonks.com/?node_id=510925 and www.perlmonks.com/?node_id=510912 22:30
22:31 ycheng joined
fglock I'm not asking for it to work - $1==undef is much easier to implement :) 22:31
22:32 r0nny joined
fglock does /$1/ never match, or matches anything? 22:33
jisom maybe equal to /<null>/?
pmichaud I'm going to guess it matches anything
TimToady probably gives you a warning on turning the undef into a "".
pmichaud or warning, yeah, that's it. :-) 22:34
fglock ok
jisom warning's don't fail ;)
ayrnieu use fatal <warnings>;
TimToady actually, it's possible that warnings in Perl 6 are, in fact, exceptions that are resumed by default.
b_jonas oh wow 22:35
fglock home &
thanks for the answers!
TimToady my pleasure.
or I guess I should say "ja nada". 22:36
fglock TimToady: de nada
TimToady gotcha.
getting my Japanese mixed in there... 22:37
ayrnieu 'je nada', and it is French.
Arathorn bunch of pedants :D
22:38 fglock left
pmichaud What does 'je nada' mean in french?!? 22:38
pmichaud looks up his french dictionary
TimToady so someone who assists a pedant is a codepedant...
pmichaud My french dictionary doesn't have 'nada' defined in it 22:39
TimToady of course not, since it means nothing.
pmichaud but not in french :-)
Arathorn presumably it's some weird hybrid pidgin phrase
the kind of thing that Young People nowadays call 'slang'
b_jonas lol. pairs with nil as the value are left out from the dictionary, just like in lua 22:42
cognominal in French, you would answer "merci" par "de rien" 22:45
pmichaud oui, que je pense
"de rien" == "It's nothing" en Anglais 22:46
cognominal yes, we are very negative people :)
pmichaud but I'm not a fluent francophone, so I wasn't sure if there was some other meaning being bantered about :-)
cognominal with such a name, Michaud, you could speak French 22:48
pmichaud yes, my great grandfather was from the french speaking part of Switzerland, and my ex-wife was from Geneva
so, I ought to know more french than I do 22:49
PerlJam pmichaud: why? you're an american! :-)
cognominal his surname sounds very French to me :)
ayrnieu He must've taken the 'no-americanization-of-surname' road to America 22:50
PerlJam cognominal: Yes, but that's a bit of history that he had nothing to do with. He's still an american.
pmichaud PerlJam: it'd be okay with me to improve the world-awareness of americans :-)
particle_ the world is flat! 22:51
ayrnieu stop lying, particle.
PerlJam particle_: It's a great big pizza!
particle_ i'm looking at a map, ayrnieu :P
PerlJam particle_++ lol
ayrnieu particle - then you should say that the map you have appears flat.
cognominal ho, only 249 Michaud for the yellow pages in Paris 22:52
ayrnieu particle - I'd note that, even as you look at that map, you should note by your perception of map that you exist in a !flat world.
cognominal I would have bet a higher number
PerlJam cognominal: how many tassin? (my mother's maiden name)
particle_ i come from flatland
pmichaud my ancestors were more rural than Paris, I suspect -- try Neuchatel
(Switzerland) 22:53
cognominal PerlJam: 10
particle_ mmm, cheese
ayrnieu particle - then you should not have said "i'm looking at a map", which suggests !flatness. You lie again!
cognominal But BooK has a bot that can get the answer for the whole France/
particle_ the map is a line :)
ayrnieu it would appear curved to you, then. 22:54
cognominal oops, perljam that was 58, he just want to spoon them to me.
ayrnieu I mean, if you've a *good* map.
cognominal pmichaud: yes, some name betray some region. there is a site to explore that. I don't remember the url
probably they know better than us in Utah :) 22:55
PerlJam pmichaud: Hey, did I tell you that one of my ancestors was from Bern? When Cyndi and I were searching for names for the twins, I got my mom's geneology chart (which goes *way* back) and noticed a Heinrich Hertzschle (or something like that) back in 1629.
jisom wonders how many Isom's in England
pmichaud PerlJam: wow, cooool
PerlJam pmichaud: I doubt I have enough swiss to gain citizenship ;-)
pmichaud perljam: oh, I probably could've done it if Isabelle and I had stayed married another two years. But yeah, you're a bit far removed 22:56
cognominal geneanet.org/partner/guide-genealog...me=michaud 22:57
PerlJam cognominal: according to my mom's genealogy, the name Tassin originated from a small "village" or "suburb" of Paris back in 980a.d. or so. Supposedly Paris has absorbed it by now. No telling if there's any trace left. 22:58
spinclad ayrnieu: you should say that the thing I seem to have that I seem to incline to call a 'map' appears to be what I seem to incline to call 'flat' 22:59
TimToady what is this "incline" thing? One more than the previous line? 23:03
spinclad it's a lie. sorry. I was being careless. 23:04
TimToady an "inclie"?
23:04 eric256 left
pmichaud one more than the previous lie 23:04
spinclad an 'ink line'
PerlJam and inclie is where you put your inc bottle.
pmichaud aka lie++
particle_ rolls down the ink line 23:05
spinclad that is, particle's map
TimToady "down"? Is that related to the inclie?
pmichaud time to zone out... be back tomorrow noonish 23:06
particle_ hey, that inclie is my mother!
spinclad 'down' is defined as where time runs slower
TimToady incday...
so if pm goes up it'll be tomorrow noonish soonish. 23:07
spinclad or nooner sooner
TimToady that leaves am to be a downer. Seems accurate...
23:08 larsen_ joined 23:17 penk joined
Arathorn hasn't an inkling of what folks are talking about :( 23:22
jisom anyone happen to know of a simple algorithm to convert a number, in this case one byte, being the H value in HSV, into RGB? 23:27
Arathorn yeah, there's a really simple one in java's awt 23:28
also one on mathworld, iirc
23:28 larsen__ joined
jisom and I'm writting in pir, so odds are I'll need to write it into pir 23:28
Arathorn well, it's just a series of averages, iirc 23:29
the magic is in the weightings
jisom no, it'd be rainbow
b_jonas one byte? just make a table of 256 colors
Arathorn i think you can get away without using any trig
jisom I'm ok with trig
Arathorn given HSV is defined as three overlapping triangular windowing functions 23:30
s/is defined/can be considered/
www.acm.org/jgt/papers/SmithLyons96/hsv_rgb.html
you have to do a few conditionals to check it doesn't go out of gamut, other than that it's just weighting. 23:31
jisom I figured I'd upgrade from 3d on a console to 3d in an image with color being the z axis 23:32
black and white's bad for eye candy 23:33
23:35 frederico_ joined
Arathorn on a console? 23:35
jisom ansi colors
:)
23:35 Southen joined
Arathorn doesn't follow how 3D and ansi colours combine 23:35
you doing 3d ascii art in parrot? 23:36
jisom colors represent the z axis
b_jonas uh.
I've once drawn magic cubes with ansi colors
jisom sort of, mapping the results of certain complex equations across a cartesian plane
Arathorn right, that makes more sense
b_jonas I had to change the palette magenta to orange for that
jisom and I'm not doing ascii anymore
too low a resolution 23:37
Arathorn had this weird mental image of programming 3D befunge or something on a text console
with a false-colour z-axis :/
jisom I think there's a doom in 3d ascii art 23:38
Arathorn uses libcaca to test vidcap devices all the time, actually 23:39
very randomly useful gimmick
jisom and they say perl has too much white noise
hack aatv or whatever to have each screen be a valid perl program 23:40
Juerd line noise.
jisom oh yeah, white noise is sound, I think
Juerd ayrnieu: Looks nice, but is that ergonomical? By the way, the example keyboard on that page looks much like an HP netserver keyboard. 23:41
ayrnieu Juerd - a trackmouse is automatically ergonomic for me :-) But I mean that you might ask them if they can make a split keyboard. 23:43
Juerd ayrnieu: Trackpoints can be ergonomical or horrible. It depends much on what you do with it. 23:47
It sucks for drawings and long web browsing sessions.
It's better than anything that's external for focussing windows, though. 23:48
(in between typing)
Same for touchpads
ayrnieu I never used it for drawing; I don't mind it for long web browsing. 23:50
Juerd Just remember this conversation should it ever begin to hurt.
In fact, don't remember this conversation, but do remember this single word: STOP :)
23:56 iblechbot joined 23:57 nirgle 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!