Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2016-03-10

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

All times shown according to UTC.

Time Nick Message
01:58 drrho joined #pdl
07:54 drrho joined #pdl
13:53 vicash left #pdl
13:54 vicash joined #pdl
13:57 gtodd kmx's  PDL::DateTime++  :-D
14:00 gtodd but I can't automatically or manually read datetime columns from a CSV file with PDL::IO::CVS  using rcvs2("/path/data.cvs" {detect_datetime => 1 } ) ;  ... I have 2D and 1D piddles confused I think
14:14 drrho joined #pdl
15:27 gtodd since when I use rcvs21d instead of rcvs2D I can place the data in a separate piddle/columns and then cat the piddles together but only if the column is in epoch time or something.  The timestamps aren't recognized if they exist in other time formats
15:37 gtodd unless I read them into their own piddle with  my ($timestano, $data) =  rcvs1D("data.csv") ;  it seems every time I use PDL for  simple spreadsheetish sorts of tasks I get bogged down in indexing and 1D versus multi-D piddles ...
15:38 gtodd it might be some kind of interfeRence from otheR environments ;-)
15:43 gtodd and the learning "curve" (which is a bit more like a steep cliff leading to a wide plateau) is always there.
15:43 gtodd :-D
15:55 gtodd PDL has great docs (the book etc.) and there are the demos; but PODs are sometimes terse in some of the packages (I think that is a fair statement?) and at times it's difficult to find a specific example / tutorial style piece of code or documentation to work from.  Of course the non-technical can muddle through because it is perl :) and just use plain perl, or a revert to a perl interface to Excel for simple tasks :-D (sometimes PDL is too much hammer for the lit
15:58 jberger gtodd: looks like you got cut off
15:58 jberger "... hammer for the lit"
15:59 gtodd little nail). :-D
15:59 gtodd it can help to pore through tests under  /t  to get a feel for things. ...
16:00 gtodd R has a publishing tradition of shipping sample data along with packages and docs (manual pages/html/pdf) which is great but sometimes makes it hard to figure out (with R) what you need to install for a particular *task*.  so wheel->reinvent
16:00 gtodd In terms of thinking about how to make PDL more broadly useful/accessible for simple tasks, I wonder if leveraging the Task:: namespace  on CRAN  could be a way to go ...  e.g. Task::PDL::PlotTimeSeries could install everything needed to read data and plot  it as a time series .. *BUT*
16:01 gtodd also include bonus data and super simple cut and paste instructions :-)
16:01 jberger gtodd: until very recently PDL was a monolith project
16:01 gtodd that appear as lovely POD and lovely README.md on  github etc.
16:01 jberger it is only just recently trying to break up into more easily consumed pieces
16:02 gtodd yeah
16:02 jberger at which point bundles make more sense
16:02 jberger note that while I hang out here I'm not very involved anymore
16:02 jberger being much more in the web world than the science world at this point
16:02 gtodd does what I'm saying make sense in terms of bundles or tasks that focus on "Doc / Task / Howto" ?
16:04 gtodd a lot of the docs exist already this would be a sort of more pedagogical approach ... where rather than a blog entry a github repo, a CPAN Task:: and a working installation of tools required to follow a long would all be possible with cpanm Task::PDL::PlotTimeSeries (say)
16:05 gtodd oh wait in my above remake I meant to write CPAN not CRAN
16:05 gtodd remark
16:05 gtodd jberger: understood :)
16:07 gtodd ... Task::PDL::WebPublishDataWithMojolicious  ;-)
16:09 jberger I have done a few cross overs on occaison
16:09 jberger :-P
16:11 gtodd heh heh ... well a PDL version of something like  Escher.jl  or the R web stack stuff has lots of fun names to choose from ... piddle -> web   ... webble wibble wepiddle ...
16:13 gtodd all instantly publishable selfhosted on a workstation/laptop or perl enable super phone  ... or on www.piddle.io  (not taken yet)
16:14 gtodd piddle as a service
16:15 gtodd widdle would be a good consistent term ... since it goes with piddle in terms of slang references etc
16:17 gtodd so we could be publishing as piddle as widdles on piddle.io
16:22 gtodd "publishing a piddle as a widdle" (after reading Task::PDL::PublichWiddle) would mean some fascinating way of extracting data and bundling it up somehow...  with requisite imagery (SVG PNG), javascript, markdown etc.
16:24 jberger ah here it was: https://gist.github.com/jb​erger/b62332a506b364117e37
16:25 jberger not very useful here but might be useful elsewhere
16:26 gtodd heh heh I don't know if this idea would be a case of perl making "the impossible improbable" or "the difficult possible" :)
16:26 gtodd ...  simple version: replaying history from the pdl2 repl and writing out/ printing images/text to files and then building a simple  mojo app to serve that directory ....
16:27 gtodd not working yet ;-)
16:27 gtodd hey that's cool
16:32 gtodd not sure it is a false memory but I vaguely recollect some combination of fatpacking , Mojo , data structures  (w/ Sereal  or Storable ?) for making SPA's in perl
16:36 gtodd (or it could've been something done completely in JavaScript (express or whatever) that I wondered might be possible in perl )
16:39 jberger you can fatpack mojo
16:39 jberger I've never done it but I remember changes being made to allow it

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