Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2014-09-01

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

All times shown according to UTC.

Time Nick Message
00:13 chm joined #pdl
00:14 chm jberger: re PDL move to github, the work was appreciated but there was no one willing to step up to sustain support for github
00:15 mohawk what support would have been needed?
00:17 chm mohawk: I think you might consider that all your assumptions are not facts.  Also, lack of a response in 10e-06 seconds does not mean you are being ignored.
00:19 mohawk i have literally no idea what you're talking about, chris
00:25 mohawk belatedly i figure that you mean my emails in April about EUD getting no response until August does not mean i'm being ignored
00:28 mohawk really starting to figure out how much performance Gimp-Perl would lose if it didn't have PDL in it anymore
00:30 mohawk also, i'm starting to see why despite the decent theoretical support PDL offers for XS code, only 5 other modules on the whole of CPAN have it, and 4 are abandoned experiments
00:35 mohawk ...and Inline-SLang was last updated in Jan 2005
02:01 jberger chm, I too am interested, what support would have been needed? I mean, after moving to github not just maintaining a mirror
02:02 jberger maybe things aren't as bad now that sf.net has pull requests, but still, the open source world has moved from sf.net to github. critical mass is there
02:26 Mithaldu chm: i feel that by refusing to discuss pdl-porters stuff here at all, you're doing pdl a disservice. You could instead do the same thing p5p has done with great success, which is to engage in *actual realtime* conversations on irc, as the topic demands, or opportunity presents itself; and then to transfer results or result summaries to the mailing list afterwards
02:27 jberger Mithaldu: I think the bigger concern is that the PDL-porters regulars are not on IRC
02:28 Mithaldu the notably important point here is that not all topics CAN be usefully discussed in emails without frustrating the ever-living shit out of every single participant in the conversation; for example conversations where one party needs to explain something, and to explain properly needs to find out what the other party knows.
02:28 Mithaldu jberger: i'm not even sure who they are, as far as i'm concerned the list is chm, you, sivoais and vicash
02:30 Mithaldu that said, i realize that what i said in the previous line might be taken as insulting, but hope that it is understood that it is the only clear way i know to express myself, being mainly used to germanic bluntness
02:38 mohawk people who read the weblog, glory in this: https://sourceforge.net/p/pdl/code/merge-requests/2
02:41 Mithaldu additionally i would like to lodge a complaint about sourceforge as a host for PDL, as they seem intent to demonstrate their ineptness about the internet, by enforcing security standards usually found in faceless enterprise corporations, cloning github guis partially without the features that make them useful, and breaking half their website in opera 12
09:10 sivoais joined #pdl
09:10 mohawk joined #pdl
09:10 jberger joined #pdl
09:10 drrho joined #pdl
09:10 pdurbin joined #pdl
09:10 vicash joined #pdl
09:10 cbmuser joined #pdl
09:10 lungching joined #pdl
09:10 ribasushi joined #pdl
09:10 ingy joined #pdl
09:10 robinsmidsrod joined #pdl
09:10 Bender2 joined #pdl
15:01 jberger joined #pdl
22:19 sivoais vicash: was my reply helpful?
22:20 sivoais regarding machine learning
23:38 vicash oh sivoais i was away from my desk. this chat session is opened at work :) and I am vpn'ing in right now
23:42 sivoais haha, nice
23:43 vicash sivoais: yes that is helpful. i am aware of libsvm but was wondering what PDL users did
23:44 vicash the application that I am writing App::financeta is now coming along well. it works on Windows/Linux/Mac and you can plot any stock whose data is available from Yahoo! and add all those technical analysis indicators (some 130 of them, not all useful) ...
23:44 vicash i am going to add a simple Rules based language that the user writes like "buy 100 at $price when $condition happens" kind of a thing. still have to figure out a way to do that ..
23:45 vicash then i want to machine learn all the parameters that the indicators use along with the rules..
23:45 vicash and add it to the application
23:45 vicash this way say you want to trade a certain stock you can do proper research on it like how they do it in the finance world
23:46 vicash internal mathematics is done using PDL and PDL::Finance::Talib
23:48 sivoais ah, nice! There's a company here in Houston that does that for energy trading (small natural language input). Their tool is written in Ruby. <http://molecule.io/tour.html>. Not sure if they have conditions.
23:49 sivoais They use <https://github.com/nathansobo/treetop>. Based on PEGs like Pegex.
23:51 sivoais I asked them if they were going to internationalise their input and they said it would be hard since their grammar is tied to English
23:52 vicash yes i plan to use pegex or lua if  a subset of lua can be back-fit to do this
23:52 vicash although i am trying to avoid english
23:53 vicash it is just a very simple grammar lthat says "buy|sell" $quantity at $price when $condition where each of these variables can be complex structures.. the aim is to parse this and generate a lambda that can then be run like sappy/apply on the PDL dataset
23:53 vicash that's why i was asking earlier about running R's apply like function in PDL to understand what is available vs what is not
23:56 * vicash will look at molecule.io

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