Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2014-05-09

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

All times shown according to UTC.

Time Nick Message
00:48 lsm joined #pdl
13:00 mohawk joined #pdl
14:15 drrho joined #pdl
15:06 lsm joined #pdl
15:17 drrho joined #pdl
16:50 rindolf joined #pdl
18:44 chm joined #pdl
18:49 chm mithaldu: any update on the freeglut fix for POGL?
18:49 Mithaldu chm, oh, actually, sure, yes
18:49 Mithaldu let me rebase that commit
18:49 rindolf Mithaldu: hi.
18:49 Mithaldu rindolf: yo :)
18:49 rindolf Mithaldu: what's up?
18:50 Mithaldu chm: you do not seem to have pushed the latest state of the opengl repo?
18:50 Mithaldu rindolf: not much, normal work day, and fussing with opengl a bit
18:50 chm which repo to/from where?
18:50 Mithaldu git://pogl.git.sourceforge.net/gitroot/pogl/pogl
18:50 Mithaldu or
18:50 Mithaldu git://git.code.sf.net/p/pogl/code
18:50 chm mithaldu: I'll take a look.
18:51 Mithaldu i got both of them
18:51 Mithaldu and my debugery fixes aren't in, as of the latest fetch
18:51 Mithaldu i see master at 1d7465da573409ce643534b98c410664b217d008
18:54 Mithaldu otoh, you didn't answer to this yet either, so i guess you haven't actually pulled that in yet? http://sourceforge.net/p/pogl/bugs/2/
18:57 Mithaldu in that case my answer is: i'm waiting with the next patch until master is fresh again :)
18:59 chm mithaldu: This is the one I see: git/git.code.sf.net/p/pogl/code
19:00 Mithaldu chm: see above
19:00 Mithaldu just the SHA should be enough to tell whether my master's the same as yours
19:01 chm I guess I didn't see the patch.  The log is the same since I last pushed.  Let me take a look at the bugs/2
19:01 chm I guess that is where the patch is?
19:01 Mithaldu it links to my github repo
19:01 Mithaldu which you can add as a remote
19:01 Mithaldu then fetch --all
19:01 Mithaldu and then either reset master to the top of debuggery, or cherry-pick and amend, however you like
19:01 Mithaldu as my bug report describes, i'm not even sure if my fix is correct :)
19:02 chm Thanks, I see the link.
19:02 Mithaldu and similarly
19:03 Mithaldu if you see anything in the gltut branch you like, you can just go ahead and cherry-pick that too, or ask me to modify it
19:03 Mithaldu (i thought you'd looked at that stuff long ago)
19:04 chm I had hoped to get the POGL2 farther along but then work/PDL/other stuff intervened...
19:04 Mithaldu is there stuff of that you have local and not pushed yet?
19:05 Mithaldu since this is still empty :) http://sourceforge.net/p/pogl/pogl2/ref/master/
19:06 chm I had factored out the GLUT stuff from POGL into its own package (trying to become GUI toolkit agnostic)
19:07 Mithaldu well push it, so others can poke at it and maybe help :)
19:07 chm But then got hung up with getting the callbacks to work.  The current POGL GLUT implementation
19:07 chm is kind of confused and I wanted to see if a clean start would help...then I ran out of time
19:08 Mithaldu ah
19:08 Mithaldu so pushing wouldn't help because you're throwing that away and haven't started a new attempt yet?
19:09 chm Right.  Although, I recently met bulk88 at the dcbpw 2014 and his XS skills are *much* greater than mine, and he uses windows
19:09 Mithaldu oh absolutely
19:09 chm Maybe I can start something soon that will give more traction...
19:09 Mithaldu and you might even understand when he explains XS, as opposed to me :)
19:09 chm I've got some gpu computing stuff that really need to be tied in to visualization.
19:10 chm The modern OpenGL is perfect for that...
19:10 Mithaldu well if you need shaders on the quick
19:10 chm Have you heard of web2project?
19:10 Mithaldu my gltut branch enables full use of vertex and pixel shaders
19:10 Mithaldu nope
19:11 chm It is a web based project management tool that I think might be a useful
19:11 chm approach for improved, asynchronous collaboration for PDL and other projects
19:11 chm One problem that comes up a lot (seems to me) is how to hand off
19:12 chm progress/tasks in a project while keeping the lessons learned and
19:12 chm knowhow.  For example, if I have a task that I've already "figured out"
19:12 Mithaldu not to rain on your parade, but i do not think OpenGL.pm does not have a big enough developer/user base to make such a tool more useful than simply moving to github
19:12 Mithaldu -not
19:13 chm The limitation of just moving to github is that more developers
19:13 chm can work on things but I haven't seen an approach for effective coordination
19:13 chm going forwards.
19:14 Mithaldu github allows collaborators to be added to a repository
19:14 Mithaldu and tickets can be assigned to people
19:14 chm As an example:  I have a task A that I've gotten 60% done
19:14 chm Now I can't work on it
19:14 chm It would be nice to put a task opportunity + context that could be
19:14 chm picked up by someone else.  I could do a short brain dump of where things are
19:15 chm which, I hope, would make things simpler to go forward.
19:15 chm I think there needs to be a little richer context for the task and plan/architecture
19:16 chm than just assignment.  Github may have that ability by customization but
19:16 chm I haven't seen it.
19:16 Mithaldu give me a sec
19:16 Mithaldu also
19:16 Mithaldu it would be nice if you hit enter after sentences, not after lines :)
19:17 chm I'm using mibbit and it doesn't wrap at all (also cuts after some number of chars)
19:17 Mithaldu ah
19:17 Mithaldu alright, i can't find a good example, right now, but
19:17 Mithaldu in github you can add labels to tickets, and it has tools for milestone management
19:18 Mithaldu so in your specific case, you're working on a ticket, so you assign yourself to it
19:18 Mithaldu that is visible in the ui
19:18 Mithaldu when you run out of time, you make a braindump comment on the ticket, which every collaborator receives per mail, then unassign yourself and add a label to it, like "up for grabs"
19:19 Mithaldu that way both direct collaborators and casual visitors see there's something up with that ticket
19:19 Mithaldu note that labels can be defined manually with any text or color
19:20 chm I'll have to think about things some more---use cases.
19:21 Mithaldu maybe also have a read through this: http://www.quora.com/GitHub/What-is-the-best-way-to-name-GitHub-issue-labels
19:21 Mithaldu just to give food for thoughts
19:22 chm Ok.  I'll take a look.  Maybe the pieces are already there---I just haven't seen it.
19:22 chm FYI, http://web2project.org is what I was mentioning earlier and the approach
19:22 Mithaldu yep, i did google it
19:22 Mithaldu gantt charts and all that stuff
19:23 chm would be to use that as a framework plus a specialized plugin to address
19:23 chm the need for context.
19:23 Mithaldu i think your main issue is simply that sourceforge is really ugh
19:23 Mithaldu it's like russian technology
19:23 Mithaldu won't fall over, but takes work to be useful
19:23 chm :-)
19:24 chm In the meantime, lacking someone fully commited to supporting PDL and the development, I'm sticking to what I know works.
19:24 chm Extra cycles are focused on development and not site-smithing...
19:25 chm Had some partial volunteers for github transition but not much use if it gets dropped on the floor after that move.
19:25 Mithaldu i don't mind that, as long as you look at my commits ;)
19:25 Mithaldu err, bug reports
19:25 Mithaldu and do stuff with them :)
19:25 Mithaldu how do you define dropped on the floor?
19:26 chm No one keeping PDL going.
19:26 Mithaldu maybe a helpful thing to do
19:27 Mithaldu would be to write a blog post on bpo asking around who uses it for what and what their wishes about it are
19:27 chm I'm not yet convinced that the convenience of github hosting is the best thing for PDL as far as stability and reliability.
19:27 Mithaldu let me put it like this:
19:28 chm One issue with github is the many-mirrors problem.  If you search for a project, you can come up with a number of repos.  Now, which one is the official one?
19:28 Mithaldu pdl/opengl on github will not magically make maintainers materialize
19:28 chm How do your users know?
19:28 Mithaldu but it will make life easier for existing maintainers, and increases the chance of contribution by others slightly
19:28 Mithaldu also, i can actually answer that question :)
19:29 chm I'm more concerned with non-developer "users" of PDL
19:29 chm How do *they* answer that question?
19:29 Mithaldu github has a page that shows a graph of repos and who forked from who: https://github.com/posabsolute/jQuery-Validation-Engine/network/members
19:29 Mithaldu and
19:29 Mithaldu if you go to this page: https://github.com/bleuscyther/jQuery-Validation-Engine
19:29 Mithaldu it says
19:30 chm ick, doesn't seem like a good solution for the naive user.
19:30 Mithaldu "forked from karol-gontarski/jQuery-Validation-Engine" at the top
19:30 Mithaldu and if you click that, it says where the next one is forked from
19:30 chm I'm trying to make PDL friendly to non-guru scientists and users.
19:30 Mithaldu i'm enumerating things
19:30 Mithaldu please keep reading
19:30 Mithaldu no thinking yet
19:30 Mithaldu only reading
19:30 Mithaldu to reinforce:
19:31 chm ???
19:31 Mithaldu if you search for a thing on github and click on a thing that isn't the original, you'll see the "forked from" link at the top, hard to miss
19:31 Mithaldu next
19:31 Mithaldu users coming from the website
19:31 Mithaldu will have a link to the proper repo
19:31 Mithaldu and users coming from CPAN
19:31 Mithaldu can be guided to the correct repo by having a link to the proper repo in META.yml
19:32 Mithaldu both sco and metacpan know how to linkify that correctly
19:32 Mithaldu see for example:
19:32 Mithaldu https://metacpan.org/release/Moose
19:32 Mithaldu left hand, repository
19:32 Mithaldu comes from: https://metacpan.org/source/ETHER/Moose-2.1205/META.yml#L1095
19:33 Mithaldu with these three ways to ensure the normal person can find the proper repo, which avenues aren't covered yet?
19:34 chm My perception is that those are all solutions to the "problem" of many, easy project forks.
19:34 chm Yes, I know that is a plus of github.
19:34 chm I'm just not prepared to be the one to support a transition to github for PDL
19:35 chm However, I am fully commited to keeping PDL going...
19:35 chm For a while there was discussion about transitioning the development to github
19:35 chm and then eventually the full project.
19:36 Mithaldu well as long as you're unwilling to say "yes, i will use github if you migrate it to that", there's not likely to be volunteers
19:36 chm A proof of concept to port the PDL web site was made---very cool work by Joel B
19:36 Mithaldu chicken/egg
19:37 chm I'm happy to use github---just want to spend my time working on PDL / POGL and using, not transitioning to github
19:37 Mithaldu ok, how about this
19:37 chm As I mentioned, I've been trying to move PDL into a truly portable and accessible numeric computation env for Perl
19:37 Mithaldu i do set it up, with an org, and repos for pdl and opengl and see what i can migrate in terms of bugs
19:38 Mithaldu and then you look at it and see if you'd like to start using it
19:38 Mithaldu and after a *reasonable* while, you make a release for both, declaring the authorative repos on CPAN via meta.yml
19:38 Mithaldu that is the crucial bit
19:39 Mithaldu moving is pointless if you don't make it possible for people to find
19:39 chm I agree that that is the crucial point.  It is pointless to do a "partial, unofficial" move.
19:41 chm I guess I'm still not convinced about github and don't want you wasting your time to set things up when I don't yet have my head around things.
19:42 chm Thanks for the good points.  It may make more sense to start PDL3 as a github project rather than
19:42 chm trying to port the PDL2 project from sf.net.  That allows comfort for the legacy stuff---reduces risk of problems
19:42 Mithaldu maybe
19:43 Mithaldu if you use the issue tracker to take notes for yourself
19:43 chm True.  That is the part I think needs some work on my end---at least.
19:43 Mithaldu if you only use it to host pdl3 and have few collaborators you won't get to use the bits that are important :)
19:44 chm I plant to try setting up a web2project approach and will see what you think.
19:44 chm True.
19:44 chm You've given me a lot to think about.
19:44 Mithaldu there's also the question of: you sure you want to spend time administrating a web app? :)
19:45 Mithaldu also, good read: http://hokietux.net/blog/blog/2013/11/05/using-the-github-issue-tracker-for-large-projects/
19:45 chm I'm more about solving the problem.  I haven't found anything that looks like a solution so trying to invent one.
19:45 chm A google search for open source project coordination / structure wasn't very helpful.
19:46 chm Maybe software folks don't have a clear idea what to do either... :-)
19:46 chm I have to get back to things here.  Thanks for the chat.  o/
19:48 Mithaldu sure
19:48 Mithaldu have a good afternoon :)
19:57 Bender1 joined #pdl
19:57 bulk88 joined #pdl
19:57 ribasushi joined #pdl
19:57 sivoais joined #pdl
19:58 Bender2 joined #pdl
19:58 lungching joined #pdl
20:01 sivoais there was also discussion of using <https://huboard.com/> in the past
20:01 sivoais here: <http://mailman.jach.hawaii.edu/pipermail//pdl-porters/2013-June/005718.html>
20:02 Mithaldu i'm not sure we have enough people to make that useful
20:02 Mithaldu that's the main thing :)
20:03 sivoais maybe that's a chicken-and-egg problem :-P
20:28 Bender2 joined #pdl
20:28 lungching joined #pdl
21:10 sivoais joined #pdl

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