Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2016-11-20

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

All times shown according to UTC.

Time Nick Message
13:06 chm joined #pdl
16:10 chm mohawk, sivoais: I've updated the ideas behind the move to PDL Next Generation
16:10 chm Please see MrUispcpu31da!R
16:11 chm q
16:15 chm joined #pdl
16:16 chm mohawk, sivoais: Oops, the URL was https://github.com/devel-chm/PDL-Tiny
16:19 chm Mithaldu: I'm working on the outline to finish the POGL2 work which I'll put on the wiki and post to pogl-devel
16:20 chm Mithaldu: I was thinking I would like to have a setup where the devel would take place at github and be mirrored at sf.net for the official versions.
16:20 sivoais chm: thanks, read through it. It looks like a good starting point. I think I'll have some more tuits by around January.
16:20 chm I believe I saw that sf.net supports that feature now
16:21 chm sivoais: good, I'm hoping to produce some first implementations to start the creative juices flowing.
16:21 chm I think it'll take some experiments before everything jells cleanly.
16:22 sivoais I'm currently extremely busy (but a good kind of busy). Would like to finish the PDL splitting and Alien::* tooling. Then the PDL NG work.
16:22 chm Keep your eyes peeled....
16:22 chm The splitting is great.  This PDL::NG would ideally, start with PDL-2.x core as the delegate for features
16:23 chm Then we could migrate to new PDL::NG functionality as it supercedes the PDL-2.x stuff.
16:23 chm I would love to see the PDLA split updated and finished.
16:23 chm I'm strapped at the moment trying to get Perl OpenGL 2 and the PDL::NG architecture started.
16:24 sivoais I'll have the time in January for sure. :-D
16:24 chm I hope to have something interesting by then.  Good luck and good holidays!
16:25 sivoais To you as well!
16:26 chm I would also like to see full Jupyter support for PDL/Perl so we can interoperate with the python-only universe
16:33 sivoais Jupyter already works, but there are a number of bells-and-whistles to add. My plan was to improve the tests before I do that. I *think* I can do that in the nearer-term, but I have to get back to work for now first! Talk to you all later.
17:02 Mithaldu chm: i can set up mirroring in the same way PDL is mirrored i think
17:02 Mithaldu also, sorry for having been silent recently, health/work issues
17:12 pdurbin sivoais: bummer that you're too busy to play data science with me: http://irclog.greptilian.com/sourcefu/2016-11-20#i_192987 :)
18:29 chm Mithaldu: Nice to hear from you.  Hope things are better.
18:29 Mithaldu still working through things, weekend wasn't great
18:29 chm Mithaldu: I was thinking to have the git flow be github -> sf.net git rather than the other way.
18:29 Mithaldu any direction is fine
18:30 chm Mithaldu: sort of what I envision for PDLA eventually.
18:30 Mithaldu though if it has automation for that, that would be even better
18:30 chm That way POGL2 could make use of Travis-CI and we would not need the hoops we have with PDL at the moment.
18:30 Mithaldu i get a lot of mails from the mirror scripts that sf git is not responding, etc.
18:30 pdurbin Is the mirroring being done for redundancy?
18:32 chm pdurbin: Re PDL, it is because at the moment PDL master git is at sf.net.
18:32 chm pdurbin: My hope was that with the new PDL::NG work we would switch to all git development with mirroring to sf.net.
18:32 chm Mithaldu: This could be demonstrated with POGL2 development.
18:34 chm The thing that sf.net has which is not available with github is more complete web page support.
18:34 chm I have an idea for development planning that I would like to try.
18:34 chm If it works, it may be a reason to use sf.net.
18:35 pdurbin chm: ah, so some of the code isn't in git right now? There's still a switch to be made?
18:35 Mithaldu chm: fwiw, this is a thing: https://pages.github.com/
18:36 chm pdurbin: What do you mean about "some of the code"?
18:37 chm pdurbin: The idea was to generate a better work flow using http://web2project.net/
18:37 pdurbin chm: you said "switch to all git development". I'm confused but that's ok. You can ignore me. :)
18:37 chm pdurbin: oh, PDL was sf.net originally
18:38 chm pdurbin: github is very popular with a low barrier to entry
18:39 pdurbin web2Project looks interesting. Who's using it already?
18:39 chm pdurbin: Adding the mirror to github allowed us to use Travis-CI and other nice things
18:39 chm pdurbin: re web2project, the idea was to enable a way to set up projects where multiple developers could
18:40 chm contribute to the same task or tasks.  One feature I was hoping for was something like a "brain dump"
18:40 chm so if someone has figured out issues or solved problems but hasn't time to code,
18:40 chm then it would be possible for another developer to get started where they left off
18:41 pdurbin I see. I guess we use GitHub issues for that. We flag it as a suggestion for someon else to pick up some day.
18:41 chm A recent example, I was planning to use GLEW for better Perl OpenGL bindings (for years but not able to proceed)
18:41 chm Then someone recently did it himself and now we have a basis to finally complete POGL2 bindings.
18:42 chm I'd like to be able to get the thoughts for others to build on
18:42 pdurbin sure, makes sense
18:42 chm pdurbin: In my experience, github issues are a bit low level and not structured to pass information along
18:43 chm I was thinking a more general project/workflow with web access so anyone could participate.
18:43 pdurbin yeah, we often link to google docs and spreadsheets from github issues
18:43 chm (Actually, this idea is another one I've been sitting on for *quite* *some* *time* :-)
18:44 chm re POGL2, if Max had seen the project idea for GLEW bindings, he/we might have gotten things done even sooner...
18:45 chm pdurbin: The hold up for getting started with web2project is I have to set things up on a local linux install
18:46 chm ...which I don't have
18:46 chm ...and is not straightforward
18:46 chm ...this, admittedly would be a simple task for someone with the actual experience.
18:46 pdurbin yeah, for us we have a roadmap but it's pretty limited to stuff we've already committed to for grant deliverables or whatever... I don't know that we have a great place to find ideas and "nice to have" stuff. stuff we dream about doing but doing have the cycles for
18:47 chm One of my goals for this Nov/Dec is to finally push the things I've been incubating on out to where others can build or contribute
18:47 pdurbin chm: have you seen other projects use web2project effectively?
18:49 chm pdurbin: No, this is my own hypothesis that with a sort of project structure we could implement a relay race open source development where things don't get stuck because someone runs out of time or transitions off of a task.  If a snapshot of their ideas/solutions could be taken, it should be easier to pick up and go again.
18:50 chm I settled on web2project because I thought the configurability and a web platform would be a good way to increase access.
18:50 chm pdurbin: My plan is to try it out on POGL2 and/or PDL-Tiny this winter.  We'll see how the experiment goes.
18:51 * pdurbin loves experiments
18:51 pdurbin chm: that's cool. Had you considered other tools?
18:52 pdurbin It seems like it's mostly a social problem. Communication and coordination problem.
18:52 chm pdurbin: Yes, the communication and coordination is important.\
18:53 chm The limitation if Issues and other bug trackers is that they don't fit passing along lessons learned, status, the how and why from solving a problem.
18:54 pdurbin yeah
18:54 chm I'm hoping that starting with a project workflow we might add some structure to support better and especially asynchronous open source development.
18:54 Mithaldu given the work i did on that branch to make it .70 i'm not we have that in the first place :P
18:54 Mithaldu *not sure
18:55 Mithaldu i.e. the work was 90% figuring out what problems he had tried to solve in the branch and how and why
18:55 chm Mithaldu: We don't have POGL2 started.  My plan for POGL 0.70 was a resting place for the "old code"
18:55 Mithaldu i am addressing *only* this bit:
18:55 Mithaldu "limitation if Issues and other bug trackers is that they don't fit passing along lessons learned"
18:55 pdurbin I feel bad for a biggish pull request (the person behind it) because I haven't been able to give it any attention.
18:56 chm Mithaldu: I've started a list of what is needed for POGL2 development and what needs be done to fold in the work from app-shadertoy
18:56 chm pdurbin: Which pull request?
18:58 chm Mithaldu: For example: back compatibility issues, regularizing function names, how to handle perl-ish stuff versus more low level stuff, ...
18:58 Mithaldu api design is something i struggle with at the best of times
18:59 Mithaldu also, let me try and say more directly what i was trying to express
18:59 Mithaldu for one, we're already pretty bad tracking what's done and why since most of it is never actually put into an issue tracker
18:59 Mithaldu for the other
19:00 Mithaldu you're underestimating issues and i'd recommend exploring it in detail to find out about things like milestone planning, categorizing, etc.
19:00 Mithaldu it's quite a bit more powerful than anything sf has
19:00 pdurbin chm: nothing to do with PDL: https://github.com/IQSS/dataverse/pull/2985
19:00 Mithaldu and the only thing it needs to be done consistently is to keep conversation entirely in it, or if done outside, post logs in there
19:01 Mithaldu i.e. the humans need to do the needful
19:01 Mithaldu pdurbin: good example :)
19:02 chm pdurbin, Mithaldu: yes, that is the direction I was heading, I was trying to figure out a way to make it discoverable.
19:03 pdurbin Sigh. I even said, "I'd be *very* interested in reviewing a pull request" - https://groups.google.com/d/msg/dataverse-community/0Jp-hhNdM38/FhyhHUw7EQAJ
19:03 * pdurbin hangs head
19:05 chm pdurbin, Mithaldu: It may be quite possible to do what I'm thinking with github tools.  If that ends up being the case, that might ultimately be a better option.
19:05 Mithaldu :)
19:05 Mithaldu what exactly do you wish to make discoverable?
19:07 chm Mithaldu: If I've "figured out" some solutions for PDL (say improved PDL::NiceSlice, or better visualization with support for multidimensional threading) I could start tasks which would have the current goal, status, problems as understood, potential solutions, for someone else to be able to run with.
19:08 Mithaldu oh yeah, absolutely
19:08 Mithaldu particularly since, if you mention an issue in another issue, the other one will automatically gain a link back
19:08 chm Mithaldu: Then if you go to the PDL devel project planner you could see what has been done and choose to continue something that I haven't gotten time to implement.  The project could even have links to github project for specific task implementations...
19:09 chm Mithaldu: I think this could be architected on top of the github toolset (similar to my expt with web2project)
19:10 Mithaldu aye
19:10 chm Mithaldu: I don't know github well enough for that and I was hoping something a bit more generic that did not require github development
19:10 Mithaldu note also particularly the custom categories you can create: https://github.com/adamkennedy/PPI/issues?page=2&q=is%3Aissue+is%3Aopen
19:11 Mithaldu ok, now you lost me again
19:11 chm Mithaldu: I think that that capability would fit in well with the higher level web2project view.
19:11 Mithaldu but maybe that is because i'm tired :)
19:12 chm It could be because I'm not clear.  I have to go at the moment but I'll try to get web2project going this week.
19:12 Mithaldu alright
19:13 Mithaldu i hope to maybe be more coherent next week
19:13 chm Hopefully, a concrete example will be more constructive, clearer and less frustrating to discuss :-)
19:13 chm Hope you feel better soon. o/
19:13 Mithaldu haha, that it would :D
19:13 Mithaldu thanks :)
19:59 chansen joined #pdl
19:59 pdurbin joined #pdl
20:00 Bender1 joined #pdl
20:12 ribasushi joined #pdl
20:41 ribasushi joined #pdl
22:21 ribasushi joined #pdl

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