Pugs t-shirts www.cafepress.com/pugscode | Pugs 6.2.9 released | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | www.geeksunite.net
Set by stevan on 15 August 2005.
01:20 sili__ is now known as sili
svnbot6 r7206 | Darren_Duncan++ | /modules/Rosetta-Incubator : added this new sub-project folder, under which a rapid rewrite of the Rosetta database access framework for Perl 6 will take place 02:36
Khisanth hrm TSa is quite confusing 04:02
putter new PIL-Run on mm2 smoke up. not bad. speed is still an issue. some tests failed due to time limits. eg, even with a separate, extended time limit, it looks like perl5.t didn't finish. 05:51
re speed, it looks like there are some low-hanging fruit in P6::C/V. todo. 05:52
good night all &
fglock++ stevan++
05:57 integral is now known as integral|ZzZzz
autrijus putter++ 06:05
autrijus pulls and takes a look
putter please disregard fruit comment. performance analysis while falling asleep is notably error prone. 06:06
&
autrijus :D 06:07
Supaplex hehe
obra heya autrijus 06:24
autrijus yo obra 06:28
dduncan greetings
fyi, tomorrow I start writing A LOT of Perl 6 code ... the whole Rosetta framework in fact, not just a tiny corner of it that is LKT 06:30
at the same time, the perl 5 version is being rewritten
obra dduncan: what do you use rosetta for?
dduncan they are both being rewritten together, with the perl 6 version being the driving one
it is a database access layer
you use it like DBI
obra dduncan: sorry. what do YOU use it for?
dduncan doing a lot of the tedious work for me when I write applications 06:31
for example, I'm making a program to sell that is useful for genealogy, among other things
I'll also be building an automator for a creative arts competition above it 06:32
the latter one automates submissions and voting
autrijus is watching a live demo of Visual Haskell
dduncan said genealogy plus program is the foundation for a new business I plan to start 06:33
autrijus can't wait until it's ported to Eclipse / hIDE
obra Visial Haskell?
Visual, even
url?
autrijus haskell.org/visualhaskell/
dduncan obra, you make PPI right?
obra no. that's alias
I make DBIx::SearchBuilder ;)
also, RT
dduncan oh 06:34
I confused you with someone I met at OSCON who did PPI
his nick was 4 letters
autrijus "adam" was 4 letters, but his nick is Alias
obra The PPI guy is Alias
dduncan maybe that's the one? 06:35
whoever it was, while I was at OSCON, on wednesday morning I messaged in this channel, asking to meet up, and said person said they were in the presenter's lounge
svnbot6 r7207 | autrijus++ | * Make MIME::Encode use multisubs as they are meant to.
dduncan I went there, and I think he said he was giving the PPI talk 06:36
Khisanth that doesn't actually sound like alias
dduncan I think it was the PPI talk ... I know it was some talk that was about a Perl topic but not Perl 6 / new stuff 06:37
Khisanth well by wednesday I assume you mean two days ago? 06:38
dduncan no, I meant wednesday August 3rd, middle of OSCON 2005
Khisanth oh that was probably him then :) 06:39
dduncan okay, my memory may be bad about the 4-letter name thing
anyway ...
Khisanth well his name is Adam
dduncan that's probably him then 06:40
on a different matter, what's the status of PIL v2 and this Meta Model 2 stuff? ...
I get the impression that what I'm making will require them 06:41
autrijus dduncan: MM2 is being deployed to the PILv1 runtime on perl5 06:42
dduncan I'm currently under the impression that, if one simply takes a fresh subversion download, does a plain invocation of the Makefile.PL etc without any special arguments, they don't get any back end using PIL yet
I could be wrong
autrijus dduncan: er, that's wrong 06:43
./pugs -BPerl5 -e 'say 123'
will compile to PIL, then compile to Perl5, then run ti
same for -BJavaScript
dduncan so the distinction is a runtime thing, not something done at make time then?
I'll have to try that then 06:44
obviously, while I'm here daily, I miss a lot when I'm not looking too closely
autrijus er, I mean -BJS 06:45
yeah, the distinction is a runtime thing
eventually the goal is to export the pugs api out to perl5 XS
so you can inline perl6 code and have them compiled to native perl5
to run alongside with your perl5 modules and share a runtime/GC/debugger 06:46
so pugs become a huge "source filter" that translates perl6 to perl5
so to speak
dduncan I C
autrijus that will require cabalization, which I'm working on
at this moment though, instead of linking in pugs, having perl5 call out with system("pugs", "-CPerl5") is probably sufficient 06:47
it will be cool to hook that up with lib/Inline/Pugs
dduncan so, just to clarify, if I want to write Perl 6 code using metamodel etc features, and I don't care about the backend, is the perl 5 one the most complete? 06:48
autrijus yes.
dduncan okay, thats what I wanted to know
autrijus but javascript is otherwise more complete on general features
dduncan thanks
I see
autrijus and unless you want to hack the metamodel, you probably want the js one
dduncan I'll look into that then 06:49
autrijus but perl5 one is coming along quite nicely
dduncan my situation is like this, I want to be able to do these things that didn't work a few months ago:
1. declare multiple classes inside one file and have the order not matter such that they just work when invoking each other
2. test a function argument to see if it is an object of a particular class 06:50
autrijus 2. I think works in various runtimes
dduncan 3. use private attributes
autrijus 1. is a parser issue -- I'll take a look
3. may work in either js or perl5, not terribly sure 06:51
dduncan I seem to recall you said #1 had to do with the parser forgetting things
autrijus yeah
dduncan now I thought that once the transition to PIL 2 was done, that parser problem would be gone
because PIL is the memory for the parser
so the parser doesn't forget things 06:52
autrijus yes, that is correct, except that part is still in its infancy
the runtimes are by now all more capable than the parser
in terms of the OO semantics they confer
now I'm getting hack time, I'll definitely remedy this
dduncan fyi, I have been learning some things by observing the progress with Pugs' design, and other projects like PPI, ... 06:54
there are numerous parallels between what the Rosetta etc needs to do and what is being done here
autrijus oh? 06:55
dduncan eg, SQL::Routine is analagous to PPI
autrijus ah indeed, you are working with SQL what we are working with Perl
dduncan they store representations of actions to perform
yes
the ::SQLParser module is like the parser
combined, the ::SQLBuilder and ::Generic, to a lesser extent Rosetta itself, are like the back-ends 06:56
autrijus yup
dduncan I think that my decision to rewrite should greatly speed up my development progress, and have something useful to show in less time
geoffb autrijus, got a few minutes? 06:57
autrijus geoffb: sure!
(I'm multiplexing with the conference so reply may be a bit intermittent)
dduncan I have been going relatively slowly over the last 2 years so far, because everything was revising or building on what was previously there, and that I was making a point of explaining everything that changed in the Changes file
geoffb I'm thinking of doing next week's blog entry on -Ofun . . . wondered what kind of conscious decisions you made to foster that optimization, if any
autrijus geoffb: oh. plenty. 06:58
geoffb do tell
dduncan but largely its my tendency to be a ... packrabbit? forget term ...
autrijus working code is fun than mere ideas
geoffb (from your kb to a big audience, with only 1 week latency!)
dduncan I've had a great difficulty in throwing things away, once I spent the time making them, so new stuff tended to just be added to it
autrijus so whenever there is a choice I do a quick-n-dirty prototyping in whatever language 06:59
and refactor as things go
dduncan in my case, the last 3 years was mostly refactoring, and little new stuff
autrijus putter's pilrun stuff -- originally crude_repl -- is a good example
geoffb nodnod
dduncan in a way, although it doesn't actually work, my previous efforts so far are like the internals of perl 5 07:00
autrijus also, it's more fun to share code with others than having to go thru an intermediary
aside from the anarchy committer model, that explains t/unspecced/
dduncan: yes, I see the resemblence
geoffb intermediary? meaning getting things approved before doing them? 07:01
autrijus yeah, or reviewed, or somehow acknowledged
dduncan hopefully I'll be able to take you as an example/inspiration and go at this "second try" from the bottom up, moving quickly and having things working almost from the start
autrijus deadlocking is least fun of all
geoffb oh yeah
autrijus and pugs development so far is remarkably bus-error free
dduncan as a side benefit, the whole process of this "re-birth" is in public SVN and has a bunch of people interested in it 07:02
autrijus safe from potential bus-errors
dduncan: right, and hyperlinks adds to the fun.
dduncan so hopefully I'll get the occasional feedback right from the start
autrijus I scan technorati and usenets and whatnot every day 07:03
two or three times
dduncan of course, as with perl 6, my re-birth version will be strongly influenced by what I did before, so most of the thinking is actually already done ... a lot of what I plan was never implemented
autrijus and try to hand out committer bit, respond to other people's musings, and generally spread the awareness
geoffb nodnod
Like Larry in the early usenet days 07:04
autrijus aye
dduncan I know I always like to just go out there and see other people talking about my work
autrijus geoffb: many of the fun came from the perl culture -- japh, obfu, golf
geoffb :-)
autrijus tolkien poetry
;)
geoffb heh
dduncan not much of that has happened yet, though, probably because what I did so far doesn't actually work
so no one could play with it
this second time should be different 07:05
geoffb dduncan, sounds like you had your second system effect the first time around. ;-)
dduncan autrijus, you had something to play with from week one, though just 'say hello', but its still something I can emulate
autrijus I think the rt.openfoundry.org invitation mechanism made a lot of difference
dduncan: yeah, I think that is critically important
I try -- "desperately" according to allison -- to get everybody else to put whatever early work into public 07:06
geoffb autrijus, what about the rt.o.o invitations? That they existed and were easy to do?
It's a good desparation
autrijus geoffb: I hacked in that feature for rt.o.o -- it's my pet favourite feature 07:07
dduncan on a side note, Rosetta has been public since 3 weeks after I started it, with over a hundred CPAN releases, though I suppose that's less public than a SVN repository
geoffb autrijus, good on ya then. :-)
dduncan I started it in mid-December 2002, and the first CPAN release was jan 5 2003
autrijus geoffb: so I'm biased, but yes, because otherwise the process of setting up a svn account (look at svn.perl.org for example) takes days
and many casual contributors won't wake for even a day
15 minutes is the attention span
geoffb dduncan, yeah, public SVN makes a huge difference, I think.
autrijus s/wake/wait/
geoffb nodnod, very true. 07:08
autrijus so I think the essence of fun, boils down to
instant gratification
geoffb You'd think it's a silly thing, but I know in my gut that automated signup is so much better.
autrijus and a sense of wonder and discovery
geoffb nodnod
autrijus chromatic calls what we've been doing as "Imagineering" 07:09
and that phrase captures the fun spirit very well.
geoffb :-)
autrijus you better start asking me something before my attention spins back to bayesian networks :) 07:10
geoffb Sorry, c&p of log to editor window
:-)
autrijus np :) 07:11
dduncan so, I'm doing this re-birth in 2 versions simultaneously, the perl 6 version starting out under Pugs' /modules dir (part of its public SVN), and the perl 5 version probably being at svn.utsl.gen.nz/trunk/Rosetta-Incubator/ (public SVN)
geoffb Go ahead and go bayesian . . . I need to think about what you've said to ask a decent followup
autrijus cool
geoffb++ # resident journalist
dduncan the perl 6 version is mean to be done in elegant perl 6, and the perl 5 version to look as much alike it as possible, done partly through the use of various third party modules 07:12
geoffb heh, and totally by accident . . . one career I never expected to start down
autrijus mumbles something about "careers defined is careers denied"
geoffb Where's that from? 07:13
autrijus er I just randomly made it up
geoffb :-)
dduncan sounds quotable though 07:15
geoffb autrijus, when you said pugs dev is bus-error free, you were referring to deadlocking (or lack thereof), right? 07:19
autrijus geoffb: yes. there's a common saying among haskell community that GHC suffers from potential bus errors 07:20
if one of the two simons get hit by a bus, we are in serious trouble
geoffb AH! Now I understand
autrijus the past month proves that pugs is much more resilient
geoffb right, although there's still unwillingness to touch things that you had talked about wanting to work on (PIL2, for example). 07:21
autrijus but that is because there is no working code for pil2 itself
geoffb I think that's partially because noone else is doing as much paper reading, conference going, and assorted compiler hacker brain-soaking as you right now
nodnod
dduncan and I thank you for it
autrijus dduncan: :-)
dduncan by the way, how can you afford to do that? 07:22
you make a lot at your job, or have a sponsor?
autrijus geoffb: but I'm glad I took my time -- after ICFP I'm having a differnt perspective
dduncan all the travelling could be expensive
geoffb yeah, I suspected as much
autrijus dduncan: I've been freelancing for 10 years; I'm trained in getting short term jobs :)
(that pays my airfare)
dduncan so your work is mainly short stints that pay a lot? 07:23
that's the kind of work I want to get
autrijus I wouldn't say a lot
but certainly enough :)
it's very high pressure though 07:25
as you could see in my past month's journal
dduncan indeed
okay, so I'm going to bed in a few minutes ... but I can still talk/listen in the mean time
geoffb yeah, some of us have been a little worried about you.
autrijus and I do stupid things -- eg. formatting RAID HDs -- when I'm burned out
dduncan I was wondering about that ... 07:26
autrijus geoffb: yeah. but I bounce back easily, thanks to ADD :)
geoffb :-)
dduncan do you ever make backups to CD/DVD etc?
geoffb (diagnosed or just obvious?)
autrijus dduncan: no, I should do that
geoffb: diagnosed _and_ obvious.
dduncan it sounded like you lost most of your stuff to the hard drive
geoffb LOL
autrijus geoffb: ADD+Bipolar means I need to minimize stress and sense of reluctance 07:27
so I can be in the blessed hyperfocus mode constantly instead of in the cursed distraction mode
geoffb Actually, for someone with such a love of well-implemented version control, I'm surprised you don't have most of your life checked into a repo
autrijus, nodnod
autrijus geoffb: but I do, except for the mails :) 07:28
geoffb gotcha.
dduncan I include my emails in my CD backups
autrijus so you can say -Ofun is kind of my survival instinct :)
dduncan currently, my email archive since 1997 is almost a megabyte in size, just text
geoffb Someone needs to write an MTA that uses a version controlled backend . . . .
dduncan that's wrong ...
geoffb autrijus, that makes sense
dduncan I meant to say that my email archive since 1997 is almost 1 gigabyte in size, all text 07:29
not including spam, which I delete
geoffb autrijus, how do you suggest that someone who fears releasing stuff that isn't "up to their own standards" -- usually set higher than they would expect of others -- fight the urge not to release until they've "just made it a little better here and there"? 07:41
autrijus geoffb: I usually ask whether I can release on behalf of them.
geoffb How did you manage to do that the first few times?
oh, interesting, and good idea
autrijus or that I'd like to talk about their work, can they please put up it somewhere linkable.
one of my most frequent uttering in both realspace and irc is: 07:42
"url?"
geoffb heh
07:43 sili is now known as sili-underscore
autrijus it's easily the key invention of Tim Bernes-Lee and I'm very grateful :) 07:43
hey brentdax.
geoffb :-)
brentdax Hey.
obra ponders sleep
autrijus how's life? how's kontent?
obra: always a good idea
geoffb obra, better to ponder it already in bed, with eyes closed . . . . :-) 07:44
But don't think too hard about it, of course.
obra autrijus: indeed. though I keep coming closer and closer to making a continuation based web framework happen in perl5
brentdax Not working on Kontent at the moment--I keep hoping the speed fairy will come by and make working on it less painful.
autrijus obra: good, because I learned here how to make it really fast.
obra: using purely eval{} and die"" 07:45
obra heh
brentdax I'm actually working on an RC4 library for Pugs, which made me realize that Perl 6 needs a way to mark data as sensitive so that the interpreter can tell the OS not to swap it out.
geoffb brentdax, how does one do that in other languages? 07:46
brentdax Good question. Let me look around a bit. 07:47
dduncan what's wrong with the OS swapping it out?
geoffb dduncan, it's harder to clean bits off a disk . . . 07:48
dduncan what's wrong with said info being on a disk?
if its sensitive, it could be stored there in an encrypted form, right? 07:49
geoffb dduncan, he's saying, don't swap the plaintext while the encryption is still in progress
(er, I think)
dduncan well, Mac OS X 10.4 encrypts all virtual memory by default, so it doesn't matter
on that system 07:50
geoffb Yes, but how well?
dduncan I don't know, but I did hear that was a new feature of 10.4 over 10.3 ... sounded like a great one to me
geoffb nodnod
dduncan of course, I only heard about it due to John Siracusa's in depth coverage of the OS on ARs
really, this is something that a secure operating system should be doing 07:51
programs shouldn't have to worry about that
s/programs/apps/
I don't know if any other operating systems do that
geoffb Years ago I went to a session in which the speaker was demoing some Perl feature or other, and he was printing $$ each run. Then he said, which OS am I running? And at least half the audience knew it was OpenBSD, because $$ was completely randomized. :-) 07:52
brentdax Don't swap the plaintext, don't swap the key.
geoffb dduncan, it's a fair bet that most security concepts exist in some form on OpenBSD, though of course no guarantees
brentdax (Although the plaintext is the user's responsibility, I think.) 07:53
geoffb But the core OBSD folks are freaks (in a very useful way)
dduncan, and what is "ARs"?
Oh Ars Technica?
dduncan yes
geoffb nodnod 07:54
dduncan I loved all of Siracusa's Mac OS X articles
geoffb One of the few general tech news sites I read
dduncan right from DP 2 in 1999 and up
geoffb :-)
dduncan I have paid for an annual ARS subscription just to support that
fyi, its $25/yr 07:55
geoffb adds that to his "When I'm donating again" list
dduncan my donation is an automated Paypal subscription, since about 3 years ago
or 4 years ago 07:56
geoffb nod
dduncan renews in february
and now nice for me to learn, some time later, that John is also a big part of the Perl community as well 07:57
not that his articles didn't mention the language every time
geoffb What does he do?
(in the Perl world)
dduncan big in a way
he writes CPAN modules, and I think he's on Perl 6 language ... have to check 07:58
geoffb Ah, he's a strong community member then.
I thought you meant he was in a place of power, like Allison's right hand or something. :-) 07:59
dduncan no, not that high up
but I confirmed it ... from my 1GB email archive, he posts a lot on p6l, and he also writes CPAN modules, such as Rose-DB 08:00
geoffb good
dduncan one thing that really strikes me about John Siracusa is that he's a strong proponent of elegant design and good user interfaces 08:01
an emphasizer of quality and good practices 08:02
he certainly makes a point of bashing Mac OS X's non-spacial Finder in every X article, citing the spacial classic Mac OS finder being easier to use 08:03
he also takes the time to explain why certain things are good or bad 08:04
and the logic seems to make sense
aka be sound
geoffb nod 08:05
dduncan good night 08:13
autrijus the Yi talk by dons is excellent -- fully dynamically reloadable editor with vi and emacs modes 08:14
I just watched him changing a few lines of Yi code, then typed :reboot, and the screen refreshed with the bug in the status bar fixed 08:15
all without losing any buffers or undo history or whatnot.
the prospect of using perl6 to script this thing is exciting :) 08:16
geoffb Mmmm, Perl 6 everywhere 08:18
autrijus, thanks for the nice link in your journal
autrijus :D 08:19
geoffb Gaah, I am resisting sleep, and I have no idea why 08:21
autrijus, you must be giving off committer bit vibes. Someone just asked for the committer bit on one of my projects (SDL_perl 1.x Resurrected), 08:46
:-) 08:47
It's been a while.
autrijus :D
spread the bits
geoffb To anyone I can. 08:48
Juerd Okay, I'm off to see feather 09:06
Wish me luck
autrijus Juerd++ # luck! 09:07
geoffb Good luck, Juerd 09:10
r0nny is there a win32 binary bundle of pugs ? 09:13
geoffb PxPerl, though I don't know how up to date it is (given I haven't used it)
r0nny want pxperl perl 5 ? 09:14
wasnt
geoffb I thought it had Pugs and Parrot in it.
r0nny well - found a win32 bin of pugs 09:15
btw - is there a vim syntax file for perl6 ?
geoffb Yes, I think it's even checked in 09:16
hold on, I'll check
util/perl6.vim maybe?
yep, that's the one 09:17
r0nny hmm
where to download ?
geoffb It's in the Pugs source tree
autrijus svn.openfoundry.org/pugs/util/perl6.vim 09:18
r0nny thx 09:21
will make live munch more easy
btw - is there a howto for threads & stuff ?
geoffb t/unspecced/ 09:22
r0nny there is no spec for perl6 threads ? 09:25
damn
geoffb We here in #perl6 live on the bleading edge 09:26
r0nny yeah
-_-
ok
now im going to do something insane 09:27
im going to write my cool irc but with multiserver support in c++
but=bot
geoffb oh the humanity 09:28
or inhumanity, in this case
r0nny hehe
as soon, as i have enough classes, it wil lbe simple
masak wow. the vim syntax file does help a lot 09:37
luqui++
nothingmuch is upset 09:48
my computer is not going to sleep on it's own anymore
geoffb nothingmuch, someone stopped by #sdlperl and picked up a committer bit there, saw your nick on the project member list, and seemed impressed. You appear to have fans. :-) 09:50
nothingmuch err, i promised to help out
who was it?
geoffb alexe
nothingmuch do i know alexe? 09:51
geoffb Who I'd never met before
geoffb shurgs
nothingmuch so
now that you've mentioned it
i really need to help out
sdlperl1 repo? 09:52
geoffb In the channel topic
nothingmuch luqui++ for blogging about unreadable papers 10:00
geoffb Oh yeah, he's totally right there. 10:04
I just don't have the attention span to read that crap
I think part of the problem is that academic papers are not yet optimized for the web way of doing things. 10:06
Aside from way too many being in PDF (*SIGH*), they all assume you're coming to them from a background in the field,
rather than assume you just came from a Google search. 10:07
"No, thank you, I don't want to have to spend 15 hours trying to read enough Wikipedia articles to understand the obtuse jargon and silly formulas strewn throughout. Yes, I'm sure it's nice and consise. Waste a few thousand bits on some extra words. They're cheap, you know." 10:08
nothingmuch i actually like PDF as a format for enhanced readability 10:11
on OSX it's fast
it's clean
it's pretty
and it prints well
as for 15 hours of wikipedia - i don't want to do that either
but i do ;-)
geoffb The only OS X box in the house has a tiny fraction of the screen real estate of my workstation . . . but my complaint is more that I can't use any tools other than a printer and a PDF reader on it. 10:12
nothingmuch over one to two years of use i've learned more from wikipedia than I have from all the encyclopedias i've read over the years and all the science channels i've watched combined
ah
in tiger the pdf reading is integrated into safari
my only complaint now is that I don't remember to save stuff as often as I should ;-) 10:13
geoffb Yeah, Wikipedia just never ceases to amaze me
pdcawley_ I find the integrated into safari thing really annoying with PDFs. 10:31
geoffb *yawn* 10:34
Time to scurry off to dream land
night, all
nothingmuch ciao
geoffb and to you 10:35
Juerd rehi 10:49
feather is back up
I have no idea what was wrong
The console cart was in use, so I had to just attempt a reboot
And that worked
The blinking of the keyboard lights suggested a kernel panic, though.
Can sameone quickly confirm that they can reach feather? 10:50
rafl Juerd: Works.
Juerd Thanks 10:53
11:09 autrijus is now known as autrijus_tw
pdcawley_ reads TSa on exceptions and wonders if 'terrorism' is the new 'Hitler' yet. 12:28
pdcawley_ is enormously glad that he won't be summarizing on Monday. 12:29
CATCH Exception { say "Why do you hate freedom?" } 12:31
autrijus pdcawley++
pdcawley_ If ever there was a candidate for the cluebat, it's that post.
rep which post 12:34
pdcawley_ The one where TSa declares that deliberately throwing an exception 'is terrorism'.
rep url?
pdcawley_ Umm... it's in the p6l Exceptuations thread. 12:35
rep :/
pdcawley_ MessageID: [email@hidden.address] 12:36
What he actually says is "BTW, I would call *intentiona* exceptions terrorism"
Which seems to miss the point that *all* exceptions are intentional.
rep hehe 12:37
pdcawley_ thinks "To hell with it!" and posts the "Why do you hate freedom?" crack to the list. 12:40
rep why do you want to destroy our way of life?
autrijus s/way of //
pdcawley_ Is that the terrorist asking the US, or the US asking the terrorist? Or both?
rep heh. 12:41
pdcawley_ stops with the politics.
autrijus I think it's terrored subjects asking the US terrorists, but that may just be me :) 12:42
autrijus also stops with the politics
kgftr|konobi theorbtwo: moo? 12:46
Limbic_Region kgftr|konobi - theorbtwo is working on a vector search engine for a contest with a Monday deadline - I would expect him to be less visible then normal 13:08
IMO, the vector search engine he is modeling is crap - but he knows what he needs more then I do
kgftr|konobi bah... he needs to get his ass out to the pub this evening
Limbic_Region when I see him at the Monastery (or castaway) - I will let him know 13:09
though from my little experience with him - he isn't much of a drinker
kgftr|konobi Limbic_Region: anyone introduced him to hilberts curve yet?
Limbic_Region: heh, with castaway at sophos, he will become one... =0)
Limbic_Region *shrug* 13:10
kgftr|konobi theorbtwo: search.cpan.org/~teejay/Math-Curve-...Hilbert.pm - check it out 13:11
Limbic_Region the approach he is using takes the cosine of the indexed document and the search vector 13:13
the problem is that it doesn't take into consideration a lot of things
kgftr|konobi nice thing about hilbert is that 2 things that are close together in 2D space will become close together in 1D space, so makes bringing up the index stupidly fast 13:15
Limbic_Region well, there are other things to consider too - word weighing, exact phrase matching, etc 13:16
and perhaps for what he needs to do to win the contest - those are all un-necessary bells and whistles
kgftr|konobi true 13:17
Class::Indexed also++
Limbic_Region I think the task for the competition is to rate which TV shows you currently like and predict which ones you might like 13:19
it was towards the end of the day and I didn't get many particulars from him 13:20
nothingmuch [email@hidden.address] 'Hi <firstname>, You Need To Act Fast To Earn Fast! ?' 14:09
Action: defer/accept/discard [dax]:
putter dduncan, autrijus: re backends: pugs is still the best all around backend. js has better non-oo, but little oo. p5 is currently weakest of the three (slowest, least coverage, most buggy, little oo). this can all change of course. perhaps even quickly. but that's my understanding of current state. 14:17
nothingmuch is getting ready to revamp BlondieD 14:19
reading a lot about Minley typing
uhm... maybe i've confused the name
putter autrijus: :)
nothingmuch oops
Hindley and Milner ;-)
so Minler typing
putter err, nothingmuch: :)
fglock__ putter: the new fails are mostly timeouts? 14:21
putter autrijus: re bus failures, I don't know that's it's been stated explicitly, so... One key contribution you make is creating a sense of high potential, of great possibilities. I could be working on pypy or ruby... but they would not dream of integrating H-M typing. And neither would pugs, without you. That excitement and motivation is fundamentally important. It just decays slower than actual commits when we think you will be back i 14:28
n a bit.
autrijus putter: sure, but the high potential was inherent in the synopses
which is also why it decays slowly. 14:29
putter fglock__: I don't have a clear idea. it might be worth running some tests individually which used to pass, but now fail. there is definitely a new failure mode, a cpu-free hang.
ingy hola friends 14:37
autrijus greetings ingy san! 14:38
ingy hi autrijus 14:39
rafl What's Luke Palmers nick on IRC again?
autrijus luqui
ingy where are you these days?
autrijus I'm in Tallinn
for ICFP -- see my journal
I'm in the "future of haskell" discussion workshop
at this moment
rafl autrijus: Thank you. 14:40
qwr lives in Tallinn ;) 14:41
ingy Ross Mayfield's wife is from ee 14:43
(my boss)
putter autrijus: yes and no. yes, if you bus collided, pugs is at a point where a couple of people could create a perl6 over a year or few. but no, it would be a much more poorer, limited and impoverished p6. both as a language, and as an implementation. the "unopened christmass present" feel of pugs, that "anything" is possible, and on a startlingly short time-scale, that p6 can be really important, not just nice... that's key to attracti 14:47
ng and keeping good motivated poeple. and I don't see anyone _else_ hanging out at ICFP, playing with simon, _sending back links and thoughts of what pugs could be_.
autrijus purrs and blushes etc. 14:48
putter sorry, big picture, you're still a potential bus error. :)
rafl Would someone please restart evalbot? 14:53
Limbic_Region would but he isn't someone he is no one 14:58
putter non?existential philosophy? ;) 15:13
Limbic_Region I didn't say "nothing" 15:14
though I have dabbled in that too 15:15
putter rafl: sorry, i don't have access to the machine on which it runs, nor am i set up to run it myself. :/
Limbic_Region: :)
Limbic_Region I would say more like borgism
rafl putter: Thought it runs on feather..
Limbic_Region I have an acct on feather but can SSH from where I am 15:16
rafl I simply don't know where the scripts are that recompile/restart it on every commit. I would do that mysef otherwise.
putter rafl: could be. don't think i ever got an account. 'locate' (if linux) to find pugs installs? watch process list for the crontab job going by? 15:20
rafl Ah, I guess I found it. 15:21
putter check evalbot source in svn for embedded filenames? ;)
great.
rafl liqui used to run it. 15:23
putter ohhh, borg-ism. took a failed wikipedia and onelook search for that to sink in. 15:25
rafl: when you figure out what to do, please document... hmm... you did check that someone didn't already? (very very fuzzy foggy recollection of iblech doing something like that... maybe) 15:26
rafl putter: I think I figured it out. It's compiling pugs by now. 15:28
putter: I checked for other docs in the repository. Nothing found. 15:29
putter k 15:30
rafl ?eval say "Hello"; 15:46
evalbot_ Hello bool::true
rafl OK, only the revision is missing now. 15:47
?eval say $?PUGS_VERSION; 15:49
evalbot_7207 Perl6 User's Golfing System, version 6.2.9, August 3, 2005 (r7207) bool::true
rafl Yay!
kolibrie rafl++
rafl Maybe someone want's to do a commit so we can see if it updates correctly. 15:50
?eval 1 while 1 15:53
evalbot_7207 (no output)
rafl ?eval "x" x 2**42 15:54
evalbot_7207 pugs: out of memory (requested 1048576 bytes)
16:07 integral|ZzZzz is now known as constant
ods15 ?eval sleep 2 16:22
evalbot_7207 Error: No compatible subroutine found: "&sleep"
ods15 ?eval $a = 1; while($a++) {}
evalbot_7207 Error: Undeclared variable: "$a"
ods15 ?eval my $a = 1; while($a++) {}
evalbot_7207 (no output)
ods15 heh 16:23
?eval my $a = 1; while($a++) { say 'a' };
evalbot_7207 a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
ods15 bah
rep .. 16:24
jql infinite loops in finite time? sweet 16:33
sili_ it bends the mind 16:35
kgftr|konobi might still be flying along... IRC has message length constraints 16:36
rafl kgftr|konobi: No. That would consume much CPU time which it doesn't currently. 16:37
kgftr|konobi rafl: "would" translates to "shouldn't"... and "shouldn't" translates to "assume" which we all know... 16:40
anyways
pub
putter Is it still the case on recent perls that method calls are > 10x more time costly than sub calls? 16:45
putter realizes he may have been making design decisions on obsolute data...
s/obsolute/obsolete/ 16:46
putter onelook's for "obsolute"...
nope 16:47
ah well. it's the way of clarity too.
pdcawley_ Oh my lord. I continue to boggle at TSa. "Exceptions shouldn't be considered just another tool."
dudley_ WTF? 16:48
pdcawley_ Well, initially he said that he called intentional exceptions 'terrorism', so I responded pointing out that there's no such thing as unintentional exceptions, and that's how he responded. 16:50
perl6-language
dudley_ I read the terrorist thing, then promptly began ignoring the thread. 16:51
pdcawley_ Yeah. I think terrorism might be the new Hitler. He' s coming dangerously close to going into my killfile, but I can't really do that what with writing the summaries and all.
putter soo... what is the current relationship between Bit and Bool? 16:52
fglock: ping? 16:54
obra Expn TSa?
pdcawley_ Thomas Sandlass 16:56
He mostly wibbles about types. Dunno if he's ever contributed any code.
fglock__ putter: Bool is an "enum" 17:12
bool::true is Bit(1) and bool::false is Bit(0) 17:13
implementation-wise, I think bool::true is a macro 17:14
putter thanks! :)
hmm... does mm2 have enums... 17:15
hmm. i would expect a bit(0) to stringify as "0", but a perlish false to stringify as "". looks like ./pugs uses "". sigh. 17:18
fglock__: how would you feel about merging P6::Value, ::Container, PrimP5 and PrimP6? i have a toy which is starting to separate a combined file into p5, mm2 p5 extracted from p6 class defs, prelude p6 for compiling to p5, and p6 for extending ./pugs. any thoughts? 17:25
the combination is rather large. but shrinking noticably. i'm still unclear on how to organize it clearly. 17:26
so it's still a "this may all just not work". but I thought i'd mention it. 17:27
fglock__ to start, how about moving into a single dir
putter hmm, good idea.
Limbic_Region fglock_ how went/goes the junction optimization? 17:28
fglock__ putter: I wrote a small example to find out what continuations and lexical scopes would look like
putter ! 17:29
fglock__ Limbic_Region: it looks like the main difference of cpan's Perl6::Junction and Q::S is that Q::S does short-circuiting
putter using the paper autrijus mentioned, or another approach?
fglock__ which paper? 17:30
putter looking...
fglock__ Limbic_Region: so it's doable
putter www.cs.brown.edu/~sk/Publications/P...tack-insp/ 17:32
www.colabti.de/irclogger/irclogger_...-09-27,Tue contains some discussion re using w p5 17:33
fglock__ putter: I'll compare with my newly invented wheel :)
Limbic_Region cool - will keep an eye on it, thanks fglock_ 17:34
putter :)
fglock__ putter: my plan was to use no stack at all - I was storing everything in a hash and using gotos 17:36
putter make sense. "hash" == "heap" 17:37
s/make/makes/
have to worry about gc yourself, yes? 17:38
fglock__ probably, yes. I wonder how p5 behaves if you have a program with thousands of gotos
putter ;) ? 17:39
fglock__ I'm not using function calls and return - only goto $h{label} 17:40
putter can't see that it would necessarily be a problem. but I haven't looked at how it's really implemented. 17:41
try! :) 17:42
oh, right. fglock, could you give me the quick tour of what's boxed and not, when Array elements are Scalars or not, and such? 17:44
fglock__ let's see... 17:45
Value.pm has 2 parts - the first half implements boxed values; the second half implements conversions of p5-p6 unboxed values 17:47
Perl6::Value::Bit::to_str() converts a p5 logic value to a p5 string, following p6 rules 17:48
putter P6V::Bit/Num/Str... is anything ever blessed into them, or are they "utility" classes. 17:49
ie, collections of subs?
fglock__ they are just collections of subs - they work on unboxed things 17:50
putter k
fglock__ Array->fetch($i) always return a Scalar
which is bound to the array element 17:51
putter so there are two levels, regular p5 str/num/etc, with some utility subs, and the mm2 classes.
fglock__ yes
putter k
internally Array stores... what?
there seemed some uncertainty in the code/comment about whether things other than ->fetch() return Scalar? 17:52
fglock__ Array.pm has 3 main parts: the MM object, the lazy array implementation, and the native array implementation 17:53
there is another object for "slice"
sorry - where? 17:54
putter Container::A being the lazy?
fglock__ yes
putter are the contents at all different in each of the three? 17:55
re where, # XXX - I think only fetch() need to return Scalar 17:57
fglock__ lazy arrays are ARRAY of (Perl6::Value::List | Object)
putter what does the usage chain look like? mm uses lazy, lazy uses Native and List, Native uses p5? 17:59
fglock__ re where: that's because assigning to a pop() doesn't make much sense anyway
putter ah
fglock__ lazy uses p5 and List 18:01
putter what uses Native?
fglock__ Native is meant to be used by the "eager" role (unwritten) 18:02
is is used to tie p5 @INC 18:03
"tieable" and "readonly" roles are implemented in Scalar.pm 18:04
"eager" and "lazy" were not written yet
putter ok, so mm Array is a wrapper around lazy C::A. is it a "thin" wrapper or does it do any conversion as requiests go by?
k
fglock__ Array processes slice() - but this could be moved into C::A 18:05
stevan hello all
fglock__ hi stevan
stevan I am giving out commiter bits if anyone is interested : rt.openfoundry.org/Foundry/Project/?Queue=423 18:06
putter hi stevan
:)
so you imported mm2? 18:07
stevan putter: no,.. this is a real-world usable version
it is based on MM2
fglock__ putter: re slice: most of the work is done by the Perl6::Slice object
putter ah C3 18:08
so lazy is full of Scalars and Lists? anything else? 18:09
nothingmuch bah
nothingmuch didn't notice there was already a window opened
TSa is becoming hard to cope with
i don't know what it is he's got against exceptions and freedom, but it's really nonsensical 18:10
fglock__ putter: that's it
putter ok...
i noticed there were a lot of is/does => { next method }. that can all go away now, right? given mm2 inheritance working 18:12
fglock__ we will need to have separate Perl6::Container::Array::Native and Perl6::Container::Array::Eager - because the Native version always force storing unboxed things 18:13
re next_method: ok 18:14
putter ok. I think that's all my questions for now. thanks! :) fglock++
kolibrie stevan: looks fun. I'm game. 18:23
stevan kolibrie: your in :) 18:25
kolibrie stevan: thanks 18:29
18:39 erg_ is now known as erg
brentdax Does pugs have an unpack() or something else I can use to split a string into pairs of characters? 18:41
Limbic_Region you could use a regex 18:42
integral .chars + a two-at-a-time loop
Limbic_Region heh
brentdax Figured as much.
Limbic_Region TIMTOWTDI is alive and well in p6
so integral - does .chars return the list of chars in one context and the count in another? 18:43
integral isn't that the subject of yet another silly thread?
Limbic_Region I posted this to the list some time ago and never got an answer - ppl have a hard time staying on topic and the thread went off on some tangent
integral I wish the easy stuff would just get decided and coded, and the hard stuff throughly ignored 18:44
Limbic_Region that's why I gave up on trying to help with the docs
actually - it is a large reason why I am not very involved at all anymore 18:45
integral nothing seems to be actually specified with rules either :-/
Limbic_Region well - the crux of the problem is that the language is still being designed concurrently with an implementation 18:46
QtPlatypus brentdax: No it doesn't. Unfortunitly pack is undefined. 18:47
Limbic_Region and while those two are trying to work together - they both have very different agendas
integral except by two groups who ostensibly communicate via wandering threads on MLs
Limbic_Region I am in violent agreement integral
integral I like the fact that the grammar is meant to be extensible using modules; but it would be even better to use that to structure the design process
err, of course that would mean development would have to implement a rules engine, and that didn't work... 18:48
well, I suppose pugs has succeded at getting the core stuff done
putter almost have a prolog based one... but set it aside to support pilrun/ P6::V/C / mm2.
Limbic_Region Pugs has succeeded in many things 18:49
and it is not nearly as bad as the picture I paint
integral the metamodel stuff is brilliant
Limbic_Region but I find it incredibly frustrating when simple questions don't get straight forward answers and spin off into endless tangents 18:50
putter one really only needs to get rules working once, anywhere. then it becomes easy to boostrap other implementations.
geoffb rafl, ping 18:51
integral you need the precedence parser bit that no one's spec'ed too
Limbic_Region integral - IIRC, .chars doesn't give you a list of chars only a count as it is intended to replace length() where you also can do .bytes etc 18:52
integral hmm, you could just write the engine in perl6 rules, use perl6 rules to parse the perl6 rules, and bootstrap via pm's stuff
putter ah, you mean for user extensions of the regexp grammar. yes.
integral ?eval $a = "abc"; say $a.chars
evalbot_7207 Error: Undeclared variable: "$a"
integral ?eval my $a = "abc"; say $a.chars 18:53
evalbot_7207 3 bool::true
integral ?eval my $a = "abc"; say join(":",$a.chars)
evalbot_7207 3 bool::true
integral ok, pugs at least doesn't do what I assumed
putter re bootstrap, until recently("historically"), pm's rules hasn't been able to parse its own grammar. still cant generate a :parsetree, but you can fake that. 18:54
Limbic_Region integral - I think Pugs is in alignment with the spec even though numerous people have requested that it support multiple return values depending on context
QtPlatypus integral: Your thinking that .chars should in an array/list context act like split('',$_)
putter then its just a matter of doing an engine. 18:55
wolverian QtPlatypus: I think .chars should return a list of aliases to the string 18:56
admittedly that is somewhat action at a distance.
QtPlatypus And likewise with .bytes?
wolverian QtPlatypus: right.
and .words
QtPlatypus endorces this product/service. 18:57
integral it just makes sense, and looks more dwimmy than split('',...)
putter I came pretty close a couple of times, but each time hit a pugsbug I couldn't work around. But a lot of that goes back half of pugs's life. Should be easier now. But still, for my latest attempt, I decided to stand someplace secure p5&prlog to get it working, and then incrementally port from there. 18:58
.chars {... want.Array ? split : length() } ? 18:59
but ($alen,$blen)=($a.chars,$b.chars)
integral want.Array? not something nice like want ~~ Array? [/me makes stuff up again]
putter i think that's waht the .t file says 19:00
integral also is this possible: ($first_char, $second_char) = $str.chars ?
wolverian given want { when Array { ... } } # would work nicely enough, although we can just implement return type MMD as well
putter (return type mmd)++ 19:01
wolverian integral: no, since .chars only returns a number. :)
integral return type or return context mmd?
brentdax Why should .chars only return a number?
putter .chars == length()
wolverian integral: I don't know. both? :)
brentdax: because that's how it is specced.
brentdax But isn't it also specced so that .bytes only returns a number? 19:02
geoffb ls
integral are they different, or are both just matched against the signature of the return continuation?
geoffb bah, wrong computer
putter length exploded into .chars/.bytes/.graphemes/.codes
geoffb ETOOMANYKEYBOARDS
wolverian integral: I don't know.
putter yes, that's spec
brentdax I'd like to see those four return pseudo-arrays into the string, which Just Happen to numify to their length. 19:03
integral are graphemes and codes the right way around there? I thought graphemes were after you turned é from two codepoints to one thingy?
putter not an ordered list
integral oops, should have been obvious =) 19:04
brentdax Smallest to largest is .bytes, .graphemes, .codes. I think there's a fourth level, but I'm not sure if it's .chars or something else.
PerlJam brentdax: that's been suggested several times with great favor from the community, nothing blessed though that I know.
brentdax Er... .bytes, .codes, .graphemes. Ack.
putter ;)
brentdax PerlJam: Isn't that what's being discussed, though.
? 19:05
fglock__ bytes/codes/graphemes thread: www.codecomments.com/archive312-200...22999.html 19:08
PerlJam wonders where .chars falls in relation toe .bytes/.codes/.graphs/.langs 19:11
s/toe/to/
QtPlatypus /me thinks that .chars == .graphs 19:13
Chars are thouse things that I see, ie graphs.
PerlJam QtPlatypus: I think that the meaning of .chars varies depending on pragmata. 19:14
geoffb QtPlatypus, I don't think that is necessarily true for all languages
QtPlatypus geoffb: Some of the wearder Indian lanauges are like that yes. 19:15
integral or maybe magical so you have one thingie that eq both "ss" and "ß"
autrijus rehi! 19:24
the conference is over :) 19:25
QtPlatypus autrijus: re
autrijus re QtPlatypus
geoffb Cool . . . how was the last day? 19:27
.oO( Long day for a last conference day ...)
 19:28
autrijus it's 10:27pm here
yeah, went to the #haskell meeting in a bar
hacked with wolfgang until the battery exhausted 19:29
nnunley It looks like you were busy.
clkao autrijus: say hi to lukhnos
nnunley!
autrijus wolfgang++ # icfp contest chD[D[D[D[D[D[D[D[Dampion
nnunley clkao!!
autrijus clkao: I don't see lukhnos here.
geoffb autrijus, might I suggest several extra battery packs?
putter QtPlatypus: re .chars == .graphs, not in unicode.
autrijus I do see ccshan
clkao autrijus: he is beside me
autrijus geoffb: they are heavy :)
clkao: oh!
clkao: lukhnos: hi!
geoffb Too bad wireless power is so . . . touchy 19:30
integral microwaving autrijus would be ... bad 19:31
nnunley Radioactive autrijus? Would he develop super powers?
PerlJam nnunley: he already *has* super-powers. 19:32
putter Tesla coils?
geoffb I was thinking more of . . . bah . . . whatshisname . . . the guy who is responsible for AC water power generation?
nnunley PerlJam: More powers are always good.
geoffb yeah, Tesla . . .
The whole "Power resonating with the fundamental frequency of the planet" thing
nnunley Mmm. Crystal planet.
putter lol
geoffb It's worth reading the story of Tesla doing this in Colorado, if you haven't already . . . the grass started to give off electric discharges, 19:33
and he eventually blew a good portion of the local power grid with the massive feedback 19:34
nnunley Nice. I remember reading some nutter's popular science book suggesting doing it again. Along with how to make a functioning UFO (which didn't)
geoffb heh 19:35
Tesla was high up on my list of "He's a genius . . . now, can we give him his own planet to mess with?" people.
nnunley Definitely a 'try it elsewhere' situation. 19:36
dduncan hello 19:41
putter ludites. :) humanity/civilization is a "give it is own planet to mess with" exercise. unavoidable. 19:42
19:42 flgr__ is now known as flgr
masak putter++ # for being a technological optimist 21:47
putter-- # also for being a technological optimist
svnbot6 r7208 | Darren_Duncan++ | /modules/Rosetta-Incubator : replaced ReadMe file with rewritten README 23:50

Powered by Cro and the Raku® Programming Language.

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

Thank you!