Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2014-12-15

| Channels | #pdl index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:36 mohawk the mailing list is a bad place for discussing stuff compared to here
01:36 mohawk in my highly-ignorable opinion
01:45 sivoais :-P
01:45 sivoais I like the mailing list for deep technical stuff, but prefer IRC for collaboration and community.
02:08 mohawk the mailing list is a very cathedral mentality
02:08 mohawk which leads to cathedral levels of agility
02:15 mohawk it has its place, but...
04:22 Mithaldu the main problem is that on the mailnig list it's not feasible to probe the knowledge of another person in detail before writing
04:22 Mithaldu so one has to decide between answering to EVERY possible knowledge combination in the other person's head, or be a german asshole
04:22 mohawk another problem is the temptation to write little essays rather than to chat
04:23 Mithaldu luckily i don't mind so much being a german asshole
04:23 mohawk Mithaldu, and for you that's not even something you need to consciously choose?
04:23 mohawk darnit
04:23 Mithaldu they're not supposed to like me, they're supposed to make good software
04:23 mohawk Mithaldu++ # self-banter
04:23 Mithaldu oh it's always a choice
04:24 Mithaldu indian person comes into #win32 and barely speaks english?
04:24 Mithaldu i'll be as friendly as to a child and patiently help them work things out
04:24 Mithaldu fully cognizant adult person with perfect english says things in an email that sound wrong? i'll say that
04:24 Mithaldu asking "did you mean that, or that" doesn't work in email
04:24 mohawk you could just say you see things differently, and spell out why
04:25 Mithaldu if that is an option i do
04:25 mohawk i think more things might be an option than you may acknowledge ;-)
04:26 Mithaldu merge --no-ff is only an option if you're dealing with actual dragons
04:26 Mithaldu i'll accept komodo warans
04:26 Mithaldu if you don't know why that is so, then you've never dealt with a dragon
04:27 Mithaldu regardless of how many years you have
04:27 mohawk dragons aren't real, Mithaldu ;-)
04:27 Mithaldu yeah wtf is this then? https://www.youtube.com/watch?v=qXhKzBAUQ_M
04:28 sivoais :-P
04:28 mohawk that would be more of a large, non-mythical reptile
04:28 mohawk unless it can fly and breathe fire and it's not in that clip?
04:28 Mithaldu (to be serious: by dragon i mean a bug you can't usefully debug via git because someone merged across things)
04:28 mohawk i'm pleased not to have had one of those
04:29 mohawk but bugs are bugs
04:29 Mithaldu fwiw, those things are actually called dragon in enlish: http://en.wikipedia.org/wiki/Komodo_dragon
04:29 mohawk i am entirely aware of that, but thanks
04:29 Mithaldu mohawk: and that's the experience thing i mentioned
04:29 Mithaldu i mean, you work on eumm
04:29 Mithaldu you've seen things
04:30 Mithaldu and i'm sure you've seen things i've never seen
04:30 mohawk i've seen things you people wouldn't believe
04:30 mohawk etc
04:30 Mithaldu but i've seen things that few other people have seen either
04:30 * sivoais dons peril-sensitive glasses
04:30 Mithaldu haha
04:30 mohawk https://www.youtube.com/watch?v=ARPCjp0ppEE
04:30 Mithaldu s/ri/r/
04:31 Mithaldu mohawk: i know :)
04:31 Mithaldu amusingly that scene fell totally flat to me
04:31 Mithaldu mainly because i watched the movie after reading these books: http://en.wikipedia.org/wiki/The_Night's_Dawn_Trilogy
04:32 mohawk i've read that, but decades after seeing blade runner ;-)
04:32 Mithaldu then i suppose you can understand what i mean :)
04:32 mohawk sivoais++
04:33 mohawk i'm sure he was partly inspired by BR
04:33 Mithaldu i'm not sure i'd call it inspired if you take a rain drop and make an ocean out of it
04:34 mohawk the atmos he made, while more space-operatic, owes something to that feel
04:34 Mithaldu that's closer :)
04:34 mohawk hence: inspired
04:35 Mithaldu i just feel the word is too small
04:35 * mohawk restrains himself from making a "that's what she said" joke
04:35 Mithaldu haha
04:35 mohawk given that we are present in a cathedral
04:36 mohawk https://www.youtube.com/watch?v=R6dm9rN6oTs
04:36 Mithaldu i guess it boils down to the fact that i think no other writer has managed to create as vast, consistent and thoughtful a world
04:36 Mithaldu every single page of his books has made me rethink a piece of reality
04:37 mohawk fascinating, captain
04:37 Mithaldu i can't get over how the first fart has too much echo in it
04:37 mohawk you may just be overthinking this
04:38 Mithaldu let me introduce you to a man i'm a fan of
04:38 Mithaldu https://www.youtube.com/watch?v=kShSq3FT36Y&list=UUq6VFHwMzcMXbuKyG7SQYIg
04:39 sivoais haha
04:42 mohawk Mithaldu, that is quality
04:42 sivoais speaking of sports, I like <https://www.youtube.com/watch?v=nI3-7rxqAlc>
04:42 Mithaldu a quality catch you surely mean
04:42 Mithaldu sivoais: i've been at parties with this guy
04:42 Mithaldu he's cool
04:42 sivoais yes, that was a quintessential catch :-P
04:42 Mithaldu also he uses entirely too many monistory to make his music
04:43 Mithaldu *monitors
04:43 sivoais hahaa
04:43 sivoais I actually have 3 monitors right here, but I'm not using the 3rd one... I should
04:44 mohawk only if you're going to make music
04:44 mohawk otherwise frankly 3 is too many and smacks of showing off
04:44 Mithaldu sivoais: you should check out his soudncloud: https://soundcloud.com/m0d
04:44 Mithaldu also you can count his monitors twice each
04:44 sivoais I need to get my MIDI keyboard from my friend... then I will
04:45 Mithaldu especially his track 404 is great
04:45 Mithaldu it goes through entirely too many music styles in the same track
04:45 sivoais nice, hadn't heard that one before
04:47 Mithaldu also this: https://soundcloud.com/m0d/quantum-of-arse
04:49 mohawk nerds
04:49 sivoais hehe, forgot about 3LN
04:49 sivoais :-P
04:50 Mithaldu sivoais: so do you go to demoscene parties?
04:50 sivoais Mithaldu: I'm in the US. Not much of the demoscene here  :-( :-(
04:50 Mithaldu there's @party!
04:50 sivoais I spectate. Sofascener.
04:50 sivoais that's in Boston, right?
04:50 Mithaldu probably?
04:50 Mithaldu srlsy, come to revision
04:51 sivoais yeah...Boston. Really far. The US is big compared to Europe.
04:51 sivoais I'd love to :-)
04:51 Mithaldu :(
04:51 sivoais once I get a job... :-P
04:52 sivoais about to post a long reply to the PDL list...
04:52 Mithaldu oic
04:53 * sivoais is still in grad school
04:56 mohawk POST IT
04:56 mohawk cathedral boy ;-)
04:56 sivoais :-D
04:56 sivoais you'll love it
04:56 mohawk that's what she said
04:56 mohawk HELL YEAH
04:57 sivoais hahahahaha
04:57 * mohawk self-5s
04:58 sivoais ok, sent!
04:58 sivoais you know what... I'll post this here too <https://gist.github.com/zmughal/4858cdf1ecd7937e594f>
04:59 sivoais it's a list of Perl projects I want to work on
04:59 Mithaldu haha, you can still list those :)
04:59 Mithaldu ( i'm deeply envious :( )
05:00 sivoais oh, this is just what I put together for someone I met
05:00 sivoais there are some more I didn't put on there because they are not as...concrete
05:00 Mithaldu also, regarding pdl tiny, i wonder if this time around we'll be able to decide column/row majority
05:00 sivoais We can do both. Eigen does both. It's a matter of abstraction.
05:01 sivoais but it's not a SMOP
05:01 sivoais At least, I don't think it is just by looking. I'll have to delve into the code.
05:02 sivoais Mithaldu, mohawk: are you reading my reply? :-)
05:03 Mithaldu looking at it
05:03 Mithaldu i'm not really qualified to comment on it honestly
05:03 Mithaldu also looking through my opengl code wondering why i can just do
05:03 Mithaldu get_dataref
05:04 Mithaldu and not need to reverse the majorality or anything
05:04 Mithaldu i guess it might be because pdl internally stores them in the same majorality opengl uses
05:04 Mithaldu and simply has a different api
05:05 mohawk sivoais, that was more than 5 minutes' worth of writing
05:05 mohawk your mission... has failed
05:05 sivoais mohawk: hahaha
05:05 sivoais yeah, I got carried away
05:06 sivoais I'm writing so much these days. I can't stop. (won't stop)
05:06 mohawk ha ha
05:06 mohawk well, got to be disciplined
05:23 sivoais eep, I realised that "you can take the mean of Kelvin, but not degrees Celsius" is wrong
05:23 mohawk yes, true
05:23 sivoais you can take the mean of degrees Celsius. You can't double it
05:24 mohawk but don't agonise
05:24 sivoais :-P
05:24 sivoais anyway, I think it might be possible to implement levels of measure with roles :-)
05:26 mohawk there ya go
17:23 gtodd joined #pdl
17:38 opkick joined #pdl
17:38 opkick [pdl] wchristian pushed 38 new commits to master: http://git.io/Ylz6xQ
17:38 opkick pdl/master d6afe70 sisyphus_: Makefile.PL - Fix doc creation fro "vendor" installs....
17:38 opkick pdl/master 1aa0d2d Chris Marshall: Update Known_problems and Release_Notes for 2.007_03
17:38 opkick pdl/master 085025a Chris Marshall: Update VERSION to 2.007_04 for development
17:38 opkick left #pdl
17:42 mohawk holy fuck
17:43 mohawk suggestion: have it message into the channel from outside, like ingy does in #inline
17:44 mohawk also, you're missing the whole "2.007_04" thing
17:52 Mithaldu mohawk: working on it
17:52 Mithaldu also, can't do the outside thing, have no op
17:52 Mithaldu someone needs to set -n
17:53 Mithaldu or convince mst to do it
18:00 opkick joined #pdl
18:00 opkick [pdl] wchristian created cpan_2.007_release at c58b3cf (+0 new commits): http://git.io/SxYKuQ
18:00 opkick left #pdl
18:00 opkick joined #pdl
18:00 opkick [pdl] wchristian deleted parallelcpu at 046edc1: http://git.io/mSBVug
18:00 opkick left #pdl
18:05 sivoais Mithaldu++ # getting PDL syncing in GitHub
18:06 Mithaldu it should be a cron job running at every xx:10 and xx:40 now
18:27 mohawk Mithaldu, when the benders get op, who has op on here?
18:28 Mithaldu i haven't the slightest clue
18:28 Mithaldu also we have channel registrations now
18:28 Mithaldu so maybe chm should be doing that and handing out cookies
18:29 mohawk so Mithaldu, do you have a commit bit on PDL's git?
18:29 Mithaldu i honestly don't know
18:29 mohawk tsk
18:30 Mithaldu i only asked for access to github so i could synch things
18:30 Mithaldu if chm did more i either didn't notice or he didn't tell me
18:30 Mithaldu also i'm too tired to brain correctly
18:42 mohawk it is sad that PDL doesn't get released more often
18:43 mohawk i am certain that PDL::Tiny will make that worse, not better
18:43 mohawk i also think it's sad that the inline-eud stuff apparently won't get incorporated
18:49 mohawk osfameron, could you please /mode #pdl -n ?
18:51 mohawk Mithaldu, please make the bot do its thing from outside!
18:51 Mithaldu word
18:52 opkick [pdl] wchristian deleted 64bit-index-support at 6519430: http://git.io/jHEfSA
18:52 Mithaldu works
19:18 jberger anyone interested in bringing the website mirror up to date (usually just the news)
19:19 jberger http://pdlporters.github.io/
19:19 jberger ^^ pure-javascript + metacpan implementation of pdl.perl.org (which is php)
19:20 mohawk i'm interested
19:20 mohawk since i gather the work has already been done
19:20 jberger yessir
19:21 mohawk what needs to be done to execute?
19:21 jberger https://github.com/PDLPorters/pdlporters.github.com
19:21 jberger its just hosted on github pages
19:21 jberger pushing to master updates it
19:21 mohawk yes
19:21 jberger that's why I made it pure+js
19:21 jberger makes it "static"
19:21 mohawk what needs doing to make that be the new PDL page?
19:22 jberger need the porters to agree (they were hesitant before) and then repoint the pdl.perl.org from sf to gh
19:22 jberger I also ported a fair amount of the wiki
19:23 jberger but I ran out of steam at some point
19:23 mohawk i think the moment is ripe to advocate it again
19:23 jberger this was before the sf internal migration when it wasn't certain what the new hosting situation would be (re: php engine etc)
19:24 jberger of course I'm a big gh fan
19:25 jberger this site does client-side calls to metacpan to render docs, rather than pdl.sf which does server side doc rendering
19:25 jberger as I recall, metacpan doesn't render the .pd file's doc :(
19:28 jberger this site also does inline MathJax: http://pdlporters.github.io/?docs=Acme::Pod::MathJax
19:29 sivoais :-D
19:29 mohawk jberger, feel like advocating this again?
19:30 jberger its much easier to advocate for if we were to decide to move to github
19:31 jberger in the uncertainty of the sf internal migration I was hoping to get some traction on that front, it didn't happen
19:31 jberger so if it didn't happen then, I assumed it wouldn't
19:31 jberger I ported the site and much of the wiki to make the move easier
19:32 jberger I think I'm done advocating at this point
19:32 * jberger has other things to fight for
19:38 jberger mohawk / sivoais: fixing this somehow would be important: https://github.com/CPAN-API/metacpan-web/issues/679
19:39 jberger after that, the only "issue" would be updating the news section
19:40 jberger which is at: https://github.com/PDLPorters/pdlporters.github.com/blob/master/content/home.html#L50-L59
19:41 jberger though it would be much better if there was another page that got dynamically included, that way it would be simpler to update
19:42 sivoais jberger: which end should that get fixed on? PDL or MetaCPAN?
19:42 jberger either
19:42 jberger run4flat thought that .pd extension should be changed to .xs.PL or .pm.PL
19:43 jberger which is semantically more correct
19:43 jberger but .pd is ancient
19:43 sivoais hmm, I see. I think fixing it in PDL would be cleaner then.
19:43 jberger I tried to argue on metacpan that they should support .pd for our historical sake, but they took the "slippery slope" opinion about adding more extentions
19:44 mohawk i fear they were right
19:44 mohawk in any case, that's ripe for build-system fixery
19:44 mohawk you rang?
19:44 sivoais the only issue would be for people browsing through old versions of PDL
19:45 sivoais scratch that, that's just indexing...
19:47 mohawk i would say getting it right for "the current version" is the important thing
19:48 mohawk when was PDL 2.007 released?
19:48 sivoais Oct 13, 2007
19:49 sivoais 12th
19:49 sivoais with a bunch of DEV releases afterwards
19:49 mohawk why does metacpan say Oct 12, 2013?
19:49 sivoais that's what I meant :-P
19:49 sivoais sorry
19:50 sivoais my fingers are being weird today
19:50 mohawk so, october 2013 = 2.007
19:50 mohawk you have fallen into... american ways of writing dates
19:50 mohawk for shame, sivoais
19:50 mohawk for shame
19:50 mohawk anyway
19:50 mohawk that's 14 months ago and counting
19:51 sivoais I just read it like it was on the site... modulo the thinko
19:51 sivoais ;-)
19:51 mohawk there have been 4 dev releases
19:51 mohawk putting aside the inline-eud stuff which there is no sign of chm actually wanting to release, why not release 2.007_03 as 2.008?
19:53 * sivoais ¯(°_o)/¯
19:53 mohawk that's a genuine question - looks like it's passing all except chorny's vortex of fail
19:54 mohawk chm, upload that shiz today
19:54 mohawk mr gorbachev - tear down this wall
19:56 jberger on a totally different topic, wanna see something silly I'm working on for fun: https://github.com/jberger/Class-Anonymous/blob/master/t/build.t
19:56 mohawk jberger, anonymous class?
19:56 jberger anonymous classes which can use lexicals as really private data
19:57 mohawk all the downsides of prototype-based OO with none of the clarity and maintainability?
19:57 jberger they don't use packages
19:57 jberger I'm not sure I'm going to recommend it's usage, its a proof of concept
19:57 jberger I've written similar things before, each time after playing with p6
19:58 sivoais jberger: hahaha. Will it support super?
19:58 jberger NOPE
19:58 sivoais :-D
19:58 mohawk the README hasn't got a "why god, why" section
19:58 mohawk it desperately needs one
19:58 jberger in the inner definition you can get the get the current method reference
19:59 jberger and then wrap that (ala before, after, around)
19:59 jberger mohawk: its only on github, if it goes to cpan (probably under Acme::) it would get one
19:59 mohawk (in case it's not obvious, i WAS being ironic about prototype-based's clarity and/or maintainability)
20:00 jberger mohawk: I now have no idea which side of this you fall on :-)
20:00 * sivoais clutches tightly on to roles
20:00 jberger sivoais: roles are trivial here
20:00 jberger it basically would be a parallel implementation as isa
20:01 jberger I envision a ::Utils with functions like around before after requires etc
20:01 mohawk jberger, i like to maintain the illusion of neutrality and balance
20:02 jberger mohawk: yeah, I know you to be a man of few strong opinions
20:02 jberger :-P
20:02 mohawk sivoais, i've asked you this before, i think - but can you point me at a very short thing that will show why roles are useful?
20:02 mohawk jberger, you're welcome
20:03 jberger mohawk: http://radar.oreilly.com/2014/01/horizontal-reuse-an-alternative-to-inheritance.html
20:03 sivoais mohawk: there was an advent calendar post just the other day that was good
20:03 sivoais that too!
20:03 sivoais <http://perladvent.org/2014/2014-12-13.html> =~ /The Role Advantage/
20:03 jberger anyway, other things are nice too, there is no need for namespace::autoclean or any of that ilk
20:04 Mithaldu mohawk: are you aware of the diamond inheritance problem?
20:04 jberger (re: autoclean, I was saying for my insane module)
20:05 sivoais the "deadly diamond of death" :-)
20:08 mohawk Mithaldu, i am
20:08 mohawk is that a problem in practice since the subclass would access  the same code via either route?
20:09 mohawk jberger, i got that
20:09 mohawk reading the links now
20:11 Mithaldu mohawk: in short: with inheritance you have a tree of things your code needs to walk through to find a function to call as method
20:11 Mithaldu and as the developer you need to be aware how it walks through
20:11 Mithaldu with roles it just straight-up imports the functions into the namespace and adds its own name to the @DOES list
20:12 Mithaldu it also makes reuse of groups of functions for easier for classes that are almost entirely different
20:16 mohawk thanks, the content looks familiar but it's sinking in as more of a useful thing now
20:17 mohawk the @DOES? roles know what can "do" them?
20:18 jberger inheritance (isa) is about sharing identity, roles (does) is about sharing functionality
20:18 mohawk that part i do get
20:18 jberger k
20:20 sivoais mohawk: you can know what roles have been rolled (heh) into an object by using ->DOES( $role_name )
20:20 mohawk oh, the inhaling class knows what roles it @DOES?
20:21 sivoais yes
20:21 sivoais but Perl's role implementations go even further. Not only can you apply a role to a class, but you can also apply a role to an object at runtime
20:22 mohawk gahh
20:22 mohawk too close to proto-based
20:22 mohawk danger, will robinson
20:22 mohawk danger
20:22 mohawk sivoais, if you're bored you may find interesting my new Alien::Gimp
20:22 sivoais I saw it! Haven't played with it yet
20:22 mohawk it's how i'll port Gimp-Perl to windows
20:22 vicash is the plan for PDL3 to use Moo::Role ?
20:23 mohawk vicash, that was my understanding from the pdlp discussion earlier in the year
20:23 vicash cool.
20:23 mohawk fortunately, we're not in any immediate danger of PDL3 development happening
20:23 sivoais an Moo-dest Proposal
20:23 sivoais :-)
20:24 vicash for one of my projects that compiles a custom DSL i converted one of my compiler backends to use Moo
20:24 mohawk that sounds painful
20:24 vicash Moo::Role was great for supporting different chips
20:24 vicash actually not really. it was necessary. i started out with Mo, then after about 8 months switched to Moo
20:24 vicash my usage of Moo::Role is very specific.
20:25 vicash If a chip performs a certain function, only then generate code for it
20:25 vicash that allowed for using Roles in a nice way..
20:25 sivoais that's really cool!
20:26 vicash so to add to the discussion, i would say that hypothetically for a JIT Moo::Role is a great idea
20:26 vicash because the JIT could detect that you're running latest Intel i7 and then use SSE4 instruction to optimize or use a GPU type thing
20:26 sivoais vicash: are using Moo::Roles to modify an AST of some sorts?
20:27 sivoais hmm, I wonder if the Rakudo / MoarVM optimizer uses roles
20:27 vicash sivoais: no. using Moo::Role to generate final code for the appropriate chip. AST generation is not done using Roles. the AST i create is generic to allow for any chip backend
20:28 sivoais ah
20:28 vicash sivoais: the compiler VIC basically parses the DSL generating Intermediate code in an AST form.. finally once the AST is complete, it then generates the appropriate chip code.. and fails if the user is trying to use a functionality that is not present
20:29 vicash that way the compiler does proper checks if the user is trying to use features that dont exist.
20:29 vicash and the whole point of this was to allow for the user to change the chip type on the cmdline and ask the compiler will the code run on another chip
20:30 vicash this allows for shortage of chips in the market to be handled by the fact that the code runs on other chips as well
20:30 vicash by chip i mean microcontroller
20:30 sivoais oh, I see. And I suppose you could eventually provide a shim for some chip features that would be pessimal, but just make it work
20:30 sivoais as a fallback
20:30 vicash well yes that is an option.. especially for things like multiplication
20:31 vicash some families provide a multiply instruction and some only provide addition
20:31 vicash but it is more useful for things like PWM, UART, SPI, I2C etc.
20:31 sivoais ah, so you can't work around those
20:31 vicash if a chip has I2C, it DOES(I2C) ... if it DOES(I2C) then code will be returned otherwise undef
20:32 vicash right now i am sticking to providing the features that the chip has and just managing those. i have not gone as far as compensating for missing features
20:32 vicash that is a good idea for the future
20:32 * sivoais hasn't done hardware in a couple years, but wants to get back into it
20:33 vicash but since microcontroller use is heavily dependent on timing, compensating a feature will never be in the same time
20:33 vicash so i would not recommend it anyway .. it is just cheaper to pay $0.20 and buy a different chip that your code is supported on
20:34 vicash i have a talk in NY.pm tomorrow on VIC . so once my slides are ready i will update it on the web and let you know
20:35 sivoais \o/
20:35 sivoais btw, this is the rakudo optimizer code <https://github.com/rakudo/rakudo/blob/nom/src/Perl6/Optimizer.nqp>
20:45 jberger btw, per-instance role application is actually just generating a new namespace, applying the roles there and reblessing
20:45 jberger part of why I tinker with anonymous class patterns
22:38 [1]mohawk joined #pdl

| Channels | #pdl index | Today | | Search | Google Search | Plain-Text | summary