Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2013-05-15

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

All times shown according to UTC.

Time Nick Message
01:57 Meiermann joined #pdl
14:37 gtodd joined #pdl
14:38 gtodd FAQ: why is Perl Data Language 2.006 newer than PDF-2.4.11 ...
14:39 run4flat gtodd, because versioning is difficult to get right
14:39 run4flat but, we didn't worry too much over it because we're moving to v3 this summer
14:39 gtodd oh :-)
14:40 gtodd I installed pdl with my OS pkg manager (BSD "pkg")  but cpan update checking scripts keep trying to grab 2.006
14:40 run4flat oh, good
14:41 run4flat that's correct
14:41 run4flat I think
14:41 gtodd even though I have pdl-2.4.11 .... guess I should use perlbrew
14:41 * run4flat double checks
14:41 run4flat yeah, 2.006 is the newest version
14:41 gtodd or somehow isolate PDL from my system perl
14:42 run4flat have you looked into local::lib?
14:42 * gtodd pkg deletes the OS package and goes for cpanm
14:42 gtodd run4flat: yeah ... use it mostly
14:42 run4flat good, just making sure you were aware of it
14:42 gtodd try to keep my base/system perl "clean" and simple
14:43 run4flat the install for PDL takes about 5-10 minutes to build and compile everything
14:43 gtodd cpanm and local::lib is how I live ... with a bit of carton and perlbrew on the side
14:43 run4flat nice
14:43 run4flat I've not looked into carton myself
14:43 run4flat gtodd, before you initiate cpanm, we should double check a couple of things
14:43 jberger perlbrew (+ local::lib) + cpanm is easy peasy awesome sause
14:44 jberger sauce
14:44 jberger o/
14:44 run4flat PDL has a number of optional build items that depend on whether it can find the libraries
14:44 gtodd run4flat: I only installed PDL as a system OS package because it is so important  ;-)
14:44 gtodd run4flat: right
14:45 run4flat so, if there's a library that you need, you should be sure that the development headers for that library are installed before installing PDL from cpan
14:45 run4flat as long as you're aware of that, and have them ready, then you're all set to install
14:45 gtodd run4flat: hmm maybe I should try restricting my cpanm based updating scripts to use local::lib somehow
14:45 jberger gtodd, I wrote a more recent install tutorial
14:45 * jberger finds
14:46 jberger http://sourceforge.net/apps/mediawiki/pd​l/index.php?title=Installing_Using_cpanm
14:46 jberger still a little old I guess
14:46 gtodd since they check @INC they try to update modules that are owned by root/system instead of the perler in question
14:46 * jberger realizes that time flies
14:47 run4flat gtodd, I thought that local::lib worked with cpanm to install the newer version into local::lib
14:47 gtodd hmm there's a script called cpan-outdated in my $PATH ...
14:47 run4flat so, yeah, it would install a newer version of PDL, but once installed, you'd have it (at least on your own account)
14:47 gtodd run4flat: it does but then I have two modules installed
14:48 jberger personally, I find local::lib a little hard to use without being under the perlbrew umbrella
14:48 run4flat and... limited hard drive space... ?
14:48 jberger but maybe I haven't tried all that hard
14:48 run4flat I think that perlbrew is best
14:48 gtodd run4flat: well, with  PDL it's probably worth it ;-)
14:49 gtodd I think there might be a root/system script that uses PDL so ... I'll update as a user and leave the old one there
14:49 pdurbin i use cpanm with local::lib. works great
14:49 run4flat but I thnk that local::lib has a low barrier to use, which is why I started down that route
14:49 run4flat I figure if gtodd knew what he was doing, he would tell me. :-)
14:50 gtodd hehe
14:51 run4flat Oh man, I just realized we didn't update the discussion subject/topic when 2.006 was released
14:51 run4flat my apologies to chm
14:51 gtodd ok nirvana will be pinto, local::lib, perlbrew together  ... and carton cpanm .... and backing everything up on github .... ;-)
14:52 run4flat ... and then waiving my hands and having it all install
14:52 gtodd :)
14:53 run4flat without having to actually type anything.
14:53 run4flat just waiving hands.
14:53 * run4flat felt it necessary to be clear on that last point
14:53 run4flat gtodd, as you're currently compiling PDL and wont' have questions, I'm gonna go get some coffee. :-)
14:53 gtodd or making two letter shell aliases to huge strings of commands that update everything and send emails ...
14:54 run4flat *won't have questions for at least 10 minutes
14:54 gtodd to your boss :-\
14:54 gtodd heh
15:02 run4flat haha
15:04 * gtodd once *again* is impressed by Prima !!
15:05 run4flat yeah, I like Prima, too
15:06 run4flat in case that wasn't obvious. :-)
15:10 gtodd it seems only barely dependent on gtk ?!?
15:11 run4flat not dependent at all
15:11 gtodd I mean on unix boxes is it X xcb or
15:11 run4flat pure X
15:11 run4flat on Windows, it's pure Windows API calls
15:11 run4flat on Mac, it uses X-quartz
15:11 gtodd hmm the file open dialog  looks gtk-ish ...
15:11 run4flat (I've contemplated leading the charge an updating it to use Cocoa)
15:11 run4flat yeah
15:12 run4flat it uses gtk *dialogs* if they're available
15:12 run4flat but can build its own
15:12 gtodd ah ok ... I don't have gnome on this system but there's probably some generic filebrowser thing
15:12 gtodd nice
15:14 gtodd on unix it looks a tad tiny small fonted olde compared to sexy gnome/kde ... but I'm sure there's config files somewhere to fix that ...
15:14 gtodd it reminds me of Irix a bit :-)
15:14 gtodd errm IRIX
15:14 run4flat interesting about font sizes. On my machine, usually the fonts are larger than usual
15:14 run4flat Prima fonts, that is
15:15 run4flat And off the top of my head, I'm not sure where that configuration file is
15:16 gtodd run4flat: Prima > perltk
15:17 run4flat gtodd, you think so?
15:17 run4flat I'm not sure, I never actually used perltk
15:17 gtodd heh :-) admittedly I don't use perltk much anymore
15:17 run4flat I've delved pretty deeply into Prima's internals, though, and think they're awesome
15:17 gtodd but Prima is pretty awesome for a small semi-secret project :-)
15:18 run4flat yeah, perfect for that kinda thing
15:18 gtodd like it compiled so fast I didn't notice .. hmm
15:19 gtodd so it is using X and xcb  (whatever the modern xlib is called) for drawing ... but is it using OpenGL or Mesa or something when it wants/needs?
15:19 run4flat no
15:19 run4flat just straight X
15:20 run4flat there is an OpenGL widget
15:20 gtodd ok it seemed to be doing a demo with OpenGL ...
15:20 gtodd ah ok
15:20 run4flat but we don't get cool OpenGL stuff, like transparency. :-(
15:20 gtodd not ... *yet* :-)
15:20 run4flat ha, the OpenGL demo was probably the PDL bindings
15:20 run4flat :-)
15:20 run4flat PDL bindings for OpenGL, that is
15:21 run4flat *yet*, indeed. :-)
15:22 gtodd hmm BSD license too  ... plus it hasn't gone down the road of cairo -> svg ..... 24 dependencies later .... -> gtk and gnome versions 1 2 and 3 ....
15:23 run4flat Yep. jberger has pushed me to write a Prima::Drawable that can output svg
15:24 run4flat (it can already output postscript, in addition to drawing to the screen)
15:24 run4flat all in good time, right?
15:24 jberger svg first, then TikZ :-)
15:24 run4flat right
15:26 run4flat Actually, the first step is to have a canvas that will draw its subcanvases
15:26 run4flat the postscript output will actually Do What I Mean for subfigures
15:27 run4flat then we'll aim for svg
15:27 run4flat and then TikZ
15:27 run4flat and then we'll conquer the world
15:28 gtodd run4flat: the last time I thought of the cool things PDL might do in the future I was thinking of web REPL (like julia-lang) and SVG canvas jscript and all the other rendering targets one could imagine via perl/cpan (probably more than R) so does Prima  abstract some of the graphics ops ?
15:29 gtodd and make rendering to surfaces (png, jpeg. OpenGL, svg)  easier to create?
15:30 gtodd I'm used to  R's approach like:     png(file="myplot.png", bg="white", otherthing="whatever")
15:36 gtodd and mostly render old fashioned postscript pgplot etc.  some R packages are appearing that seem to use cairo for rendering to lots of pretty things ... but I want to see what Prima can do from PDL ...  without TikZ for now ;-)
15:36 gtodd this may mean I will have to join mailing lists and make a fool out of myself but ...
15:43 sivoais gtodd: I'm working on it! I talked to you about it last time. <https://github.com/zmughal/devel-repl-web>
15:43 sivoais gtodd: Right now, I'm trying to add completion support to the jq-console widget that I'm using.
15:44 gtodd sivoais: yeah I remember :-)
15:45 sivoais Yeah, I'm free right now, so I've been working on it furiously. :-D
15:45 gtodd wow you have made progress!
15:46 sivoais I really have Devel::REPL to thank. I hook into their plugin system and make a mock terminal
15:48 gtodd it is pretty awesome
15:48 run4flat gtodd, Prima can act as a generic drawing backend
15:49 run4flat all using PDL::Graphics::Prima, that is
15:49 run4flat but if you're serving images to a browser, it'd probably be easier to use PDL::Graphics::Gnuplot
15:49 run4flat or another, more traditional plotting package
15:49 gtodd Devel::REPL is another matt trout creation and for that reason alone I feel we should all go along with the Pumpkin Perl idea :)
15:49 run4flat haha
15:52 gtodd run4flat: cool so there is nothing stopping someone adding Prima::CORE::WebCanvas or something web-ish to Prima itself ??
15:53 run4flat It'd be called Prima::Drawable::WebCanvas, but yeah, there's nothing stopping this
15:53 run4flat And if that exists, then I could update PDL::Graphics::Prima to work with it
15:54 run4flat (which would involve changing, like, two lines)
15:56 sivoais I'll keep that in mind. I want my front-end code to be reusable (at least, in part). :-)
15:59 sivoais I'm learning about CoffeeScript and it is surprisingly Perlish via Ruby.
16:05 gtodd yeehaw!
16:05 gtodd sivoais: why the like to longtailvideo  in DESIGN.md ?
16:07 gtodd sivoais: do you work there ? ;-) or maybe the URL supposed to point to videos of deve-repl-web in action !
16:08 sivoais gtodd: I thought it would be a cool demo to have something retrieve URLs and have it be rendered as a video player. Actually, that could probably be done better with oEmbed.
16:10 gtodd hah cool
16:12 sivoais further down the road, I want these tools to be part of a debugger, but I'm not going to tackle that just yet
16:22 gtodd sivoais: before long you are going to make it possible to install a personal workalike CKAN from CPAN
16:25 sivoais hah, are you reading my mind...? :-) I'd love to make working with datasets that easy. It would probably go half-way to implementing Wolfram Alpha, but I don't think that would be a bad thing.
16:26 sivoais I've got pushing the existing CKAN Perl client to CPAN on my Play Perl todo list
16:30 run4flat sivoais, gtodd, how does ckan compare to http://thedata.org/ ?
16:30 run4flat pdurbin works on http://thedata.org/, so I'm curious. :-)
16:35 gtodd sivoais should know ckan pretty well :-)  ... I was interested a few years ago in how tools coming out of  the whole big field of opendata were going to impact collaborative research so was advocating exploring/developing some consumer friendly tools or working with folks who were doing that ... a lot has happened since but I've been out of the loop.
16:36 sivoais run4flat: yeah, so, CKAN is a platform. You can download it and run an instance of your own with specific data. Lots of local-governments are trying to setup their own instances so people can get at that data. I'm not sure about how they are internally different, but they seem to have slightly different goals.
16:37 sivoais it looks like http://thedata.org/ has better support for the traditional information/library science tools
16:38 run4flat Hmm, interesting. My impression of thedata.org is that is was slanted towards scientists and institutions
16:38 run4flat but it sounds like ckan is more for government data
16:38 run4flat those are slightly different use cases
16:38 sivoais citations for versions! :-) Obviously important in a research setting
16:38 run4flat cool
16:38 gtodd pdurbin: I was working in the public sector and IT is dominated by operational concerns (MSCEs and Service Packs) so they tend to not see this kind of thing coming until it's sort of blindingly obvious :)
16:39 gtodd sivoais: hm I think I will try your perl CKAN client?!? what?!?
16:40 gtodd sivoais: hmm I must've noticed that last time I was in this channel :-)
16:41 sivoais It's not mine, but it is missing tests and a Makefile <https://github.com/lukec/net-ckan> (and a Changelog, etc.)
16:41 sivoais it needs to be CPANized
16:41 gtodd ok good ...
16:42 gtodd otherwise you would be a superhuman mindreader
16:42 gtodd not on CPAN = "beta"  ;)
17:01 gtodd are there any Prima devels here?  I couldn't  Prima to build/install from CPAN without using an OS patch (the FreeBSD ports/pkg infrastructure)  to img/codec_ungif.c
17:02 gtodd then again PDL-2.006 seems a tad buggy too compared to 2.4.15 (errm the older version) :-D
17:03 run4flat gtodd, I know a little bit, but Dmitry---the brains behind Prima---doesn't hang out on IRC
17:04 gtodd ok I will look somewhere to file a bug
17:04 run4flat cpan bug tracker works
17:04 liddle_piddle_bot http://p3rl.org/bug tracker works
17:04 run4flat haha
17:04 run4flat cpan Prima
17:04 liddle_piddle_bot http://p3rl.org/Prima
17:04 gtodd hmm ok PDL-2.006 errors are all due to the much hated (by me) readline ... :-\
17:04 run4flat You can go here and follow the bug reporing link: https://metacpan.org/module/Prima
17:05 run4flat gtodd, really?
17:05 run4flat I hadn't followed the latest release
17:05 gtodd ok that's a strong word ... really I just dislike it a lot (esp. in sysadmin mode)
17:05 gtodd perl should just use libedit or linenoise.c :-)
17:05 run4flat bugs for PDL can go on sourceforge.net
17:05 run4flat or, you can tell me and I can pass them along
17:06 run4flat As for libedit or libnoise, I hadn't heard of them.
17:06 gtodd rakudo/perl6 seems like it is going to depend on readline :-|
17:06 gtodd run4flat: just my pet peeve ...
17:06 run4flat If they're useful, we can wrap them in Alien modules and use them for PDL3's readline
17:08 gtodd sorry needless ran ... I don't understand (especially in BSD world) why gnu-readline is so welded into everything .. RMS tried gcc, HURD,  emacs but achieved world domination with readline  :-)
17:09 * pdurbin looks at ckan - The open source data portal software - http://ckan.org
17:10 gtodd pdurbin: neat eh?
17:12 pdurbin gtodd: very interesting. I dropped a link at http://irclog.iq.harvard.edu/dvn/2013-05-15#i_2136
17:12 pdurbin gtodd: what does CKAN stand for?
17:13 * run4flat was wondering the same
17:13 sivoais Comprehensive Knowledge Archive Network
17:13 pdurbin ah. thanks
17:14 sivoais It's from the people at the Open Knowledge Foundation
17:14 pdurbin I see it here: http://en.wikipedia.org/wiki/CKAN
17:14 sivoais they're on freenode's #ckan and #okfn respectively
17:15 gtodd yeah
17:16 pdurbin seems like our team is aware of it: https://redmine.hmdc.harvard.edu/issues/2757
17:17 pdurbin I remember looking at this screenshot: https://redmine.hmdc.harvard.edu/at​tachments/515/proposed_workflow.png
17:17 pdurbin but the term CKAN didn't stick with me
17:18 gtodd pdurbin: I used to work with a lot "open knowledge" networks 5-10 years ago in the field of IT for development  ... CPAN was something I often referenced :-)  since practically nobody was actually in IT the metaphor didn't really work :)
17:18 gtodd pdurbin: if you cycle through the variations of C*AN some work some don't :-D
17:19 pdurbin 'round here people talk about CRAN a lot :)
17:19 gtodd :)
17:22 gtodd run4flat: http://pastebin.com/u0yLLB50  <-- here's the sort of error I mean ... will fiddle more before reporting a bug since readline bugs are sort of by definition never really bugs :)
17:22 run4flat thanks
17:22 * run4flat checks
17:22 gtodd liblinenoise is so small and does 75% of what readline does I wish it was the default for $EVERYTHING but readline has what developers need I guess
17:22 run4flat oh dear god
17:23 gtodd hehe
17:23 run4flat I hate stack traces
17:23 gtodd see it's not a real bug :-)
17:23 run4flat Oh, it's a problem with *MY* library
17:23 * run4flat is now much more concerned
17:24 run4flat gtodd, you have submitted this bug to the perfect place. :-)
17:24 gtodd hehehe
17:24 run4flat oh sheesh, no, that's not a real error
17:25 run4flat it has to do with proper error catching and rethrowing
17:25 run4flat the cleanest way to break out of the Prima event loop is to throw an error
17:25 run4flat so I throw the error "time to exit the event loop"
17:25 run4flat which is *supposed* to be caught
17:26 run4flat I'll see if I can reproduce and fix this on my machine
17:26 gtodd ok with readline it's often something like error leading to  -> error -> ... -> ... -> ... -> ... A pronouncement made by RMS in 1989
17:26 gtodd so now you are giving me hope!
17:27 run4flat haha
17:27 run4flat I shall look into this, but first, lunch!
17:28 run4flat in the meantime, you should read the docs for PDL::Graphics::Prima::Simple
17:28 gtodd it happens right when I click on finish
17:28 run4flat those should give you a fine starting point
17:28 run4flat ah, good to know
17:28 gtodd from running "demo prima"
17:28 run4flat probably a race condition
17:29 run4flat or, an out-of-order problem
17:30 gtodd ok ...  by the time you finish lunch I will not have figured out how to run pdl in the perl debugger :)
17:34 run4flat gtodd, give up hope. You can't run PDL in the perl debugger
17:34 run4flat and, generally, the plotting software shouldn't give this problem
17:34 run4flat only the demo
17:35 run4flat unless... you're trying to debug it for me
17:35 run4flat which is an awfully nice gesture. :-)
17:39 gtodd hah
18:06 jberger ETOOMUCHBACKLOG
18:08 run4flat hahaha!
18:10 jberger just finished an noontime lifting session, time for lunch for me
18:10 jberger I got up to repping 540 on leg press! :D
18:10 run4flat wow.
18:10 run4flat jberger++
18:10 run4flat hehe
18:10 jberger my old max is in reach! I did one lift at 720 once in HS
18:11 jberger I hope to reach and possibly rep it by the end of the summer
18:11 jberger run4flat runs, jberger lifts
18:12 run4flat alas, I never hit 4min in the mile
18:12 run4flat best was 4:09
18:13 jberger I thought runners peak closer to your age now? you could still get it
18:18 run4flat ha
18:18 run4flat runners who consistently train tend to hit their peak in their upper 20s, early 30s
18:19 run4flat and longer distances typically have slightly later peak ages
18:20 run4flat and, I trained awfully hard in college
18:34 jberger run4flat, sometimes I wish pdl development could be more like #mojo
18:35 jberger 4.0 keeps getting pushed back because there are too MANY things changing, not waiting while the few things slowly sort out
18:35 jberger this is why I want I a small core
18:35 jberger understandible, testable, changeable
18:36 jberger <3
18:36 jberger *small core for PDL
18:36 jberger #mojo development is exciting
19:35 gtodd <3
19:47 run4flat jberger, 4.0?
19:47 run4flat don't you mean 3.0?
19:48 * run4flat hasn't followed the mailing list for a little while
21:18 jberger Mojolicious 4.0
21:18 jberger which JUST came out
21:20 run4flat ah, I see
22:22 jberger have PDL people seen this? http://stackoverflow.com/questions/​16533394/how-can-i-tell-a-cpan-inst​allation-of-pdl-to-use-my-fftw-path
22:28 run4flat no
22:29 run4flat almost certainly he has a version of FFTW that is too new
22:29 run4flat :-(
22:31 jberger I just sent it to the list too
22:31 run4flat as he fixed the problem it clearly is not version skew, as I suspected
22:31 jberger FFTW really REALLY should be a CPAN module and not core PDL
22:31 run4flat as for makepl_arg, I've never heard of it
22:32 run4flat yes, and it should have an Alien:: module, too. :-)
22:32 * run4flat adds it to the list
22:32 * run4flat realizes it was already on the list
22:32 pdurbin heh
22:33 jberger if PDL goes to a Moo(se) arrangement, then these CPAN modules (and other functionality) can all be roles applied at runtime
22:34 run4flat yes
22:34 run4flat that is essentially what we do right now
22:34 run4flat jberger, did you see my idea about namespacing PDL methods?
22:35 jberger right, but because its all done by hand, it makes it hard to do by external modules
22:35 run4flat I can dig up the #pdl log if not
22:35 jberger no?
22:35 jberger please do
22:35 run4flat I threw it out there Monday or Tuesday
22:35 * run4flat digs
22:36 run4flat starts here: http://irclog.perlgeek.de/pdl/2013-05-14#i_7067518
22:39 run4flat Or, to continue along that line of reasoning, if we require Perl 5.10, we can use lexical pragmas
22:39 run4flat "use PDL::FFT" could add "PDL::FFT" to a list of packages in the hints hash string
22:40 run4flat We could play with AUTOLOAD to have PDL's method resolution examine the contents of said hints hash
22:40 run4flat so that PDL::FFT's methods would only be available in the given lexical scope
22:41 run4flat side note: anybody remember the name of the pastebot?
22:42 perigrin shadowpaste?
22:42 shadowpaste joined #pdl
22:42 run4flat perigrin++
22:43 run4flat thanks
22:43 run4flat the channel's topic is once again accurate
22:43 perigrin heh
22:46 run4flat pdurbin, want to do me a favor?
22:46 pdurbin maybe
22:46 run4flat I think I found a fix for the demo crash
22:46 run4flat it involves changing one line in the PDL tarball
22:46 pdurbin what demo crash?
22:46 run4flat agh, never mind
22:47 run4flat ETOOMANYCHANNELMEMBERS
22:47 run4flat I meant to chat with gtodd
22:48 run4flat gtodd, do YOU want to do me a favor?
22:49 run4flat OK, clearly it's time for me to head home
22:49 run4flat o/
22:49 run4flat have a good night everyone
22:52 jberger o/
22:52 jberger will consider this stuff

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