Camelia, the Perl 6 bug

IRC log for #bioperl, 2013-04-05

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

All times shown according to UTC.

Time Nick Message
08:01 bbb joined #bioperl
08:12 aleimba joined #bioperl
09:05 aleimba joined #bioperl
12:18 looper joined #bioperl
12:37 carandraug joined #bioperl
12:37 carandraug joined #bioperl
13:46 carandraug joined #bioperl
14:38 denisboyun joined #bioperl
15:02 looper joined #bioperl
15:21 pyrimidine joined #bioperl
15:36 carandraug joined #bioperl
15:36 carandraug joined #bioperl
15:36 carandraug pyrimidine, hi!
16:06 pyrimidine howdy
16:08 carandraug pyrimidine, so what's the story with making new releases?
16:08 pyrimidine well...
16:09 pyrimidine the issue is, should we go through the effort and push out a 1.6.x release
16:09 pyrimidine or stay the core on master and do incremental 1.7.x
16:09 carandraug should I keep on splitting the repo. What path should I take?
16:09 pyrimidine split out on master
16:10 carandraug pyrimidine, I'd forget about a 1.6.x release. Seems that people who do use it, build it from github anyway
16:10 pyrimidine right, and they can simply pull from 'v1'
16:11 carandraug pyrimidine, and the ones who don't... well, let them contribute for the release. Are there that many changes that make it worth it?
16:11 pyrimidine not really.  mostly bug fixes.
16:11 carandraug pyrimidine, I'd vote on make small 1.70 releases that have very little changes. these can be starting point for 2.00 releases
16:11 pyrimidine carandraug: yes, I agree
16:12 pyrimidine carandraug: you should try for getting some funds for this
16:12 pyrimidine TPF, or GSoC
16:12 carandraug so when making the split, should it be done from the core or the outside?
16:12 pyrimidine outside?
16:12 carandraug pyrimidine, I'll apply for GSoC if we get accepted
16:13 pyrimidine ok
16:13 carandraug pyrimidine, yes. Should we start splitting from the core (bio::Root, Bio::Seq, etc), or the outside (Bio::Cluster, Bio::Biblio, etc)
16:14 pyrimidine I think outside is easiest
16:14 pyrimidine there will be a point where it will become more painful to split things
16:14 carandraug pyrimidine, it's very easy to get circular dependencies starting from the outside. I lack a well defined plan for the split
16:15 carandraug If the plan is good, I can make a 1.70 release of many at the same time, and there's no dependencies problems
16:15 pyrimidine right
16:16 carandraug for example, Bio::Location and Bio::Range, do they make sense to be split from a Bio-Seq distribution?
16:16 pyrimidine I do think there will be a point where it becomes less useful to split out code
16:16 pyrimidine right, exactly
16:16 pyrimidine but it may make sense to split those out from the perspective of changing how they work internally
16:16 carandraug should we have a Bio-Seq distribution with all the Bio::Seq::* modules, or have a BioPerl distribution that has only role classes, and have many small distributions for each type
16:17 pyrimidine Bio::Location and Bio::Range for instance are a real sore spot
16:17 carandraug when I was looking at Bio::Cluster, there was one that is for a cluster of Bio::Seq objects. Should it go in a Bio-Cluster distribution or part of Bio-Seq?
16:17 pyrimidine Bio::Cluster would have a Bio::Seq dependeny
16:18 pyrimidine core should be as lean as possible, very few to no dependencies
16:18 carandraug pyrimidine, no. Only the module that deals with clusters of Bio::Seq depend on that.
16:18 pyrimidine probably consist mainly of roles in the future (interfaces for now I suppose)
16:18 pyrimidine ah
16:19 carandraug pyrimidine, yeah, talking abuot design for BioPerl 2.00, a couple of people on #perl offered to help in discussing design
16:19 pyrimidine that would be good
16:19 pyrimidine genehack and deafferret: still plans to do something on the state of bioperl at YAPC?
16:20 pyrimidine this might be worth bringingg up
16:20 pyrimidine *bringing
16:20 carandraug pyrimidine, so about Bio-Cluster, does it make sense to have a separate distribution? Should a BioPerl core distribution have the Bio::Cluster interface module, and leave the actual Bio::Cluster::* modules up to the distributions of what is being clustered?
16:20 genehack pyrimidine: deafferret submitted a talk. i don't think they've announced acceptances yet
16:20 pyrimidine genehack: ok
16:20 pyrimidine carandraug: interface in core, modules elsewhere
16:20 genehack they did say ( # of submissions ) > ( # of slots )
16:21 carandraug pyrimidine, so Bio-core is different from bio-root?
16:22 carandraug I was looking at the stuff in Bio-Root and I'm not sure it makesa lot of sense nowadays
16:22 pyrimidine carandraug: yes
16:22 pyrimidine and agree, it doesn't
16:22 pyrimidine the only point for it was to have common core methods available to all classes
16:22 pyrimidine that does make some sense
16:22 carandraug pyrimidine, so with v2.00, there would be no Bio::Root stuff
16:23 pyrimidine I think so, yes
16:23 pyrimidine I think it adds a lot of (unnecessary) weight
16:23 pyrimidine and most (all?) of the functionality could be reimplemented as exportable functions
16:24 carandraug yes, some things there make sense but they could be implemented differently. It seemed a lot of code that is already implemented outside Bio::*
16:24 pyrimidine only things that make sense to have globally is logging/debugging (but again could be something exportable)
16:25 carandraug pyrimidine, ok. So what's the plan for splitting. How should I go on with that?
16:26 carandraug 'cos I still don't know a big part of the code. The more I see , the more dubious I get about some of the splits
16:26 pyrimidine any specific ones?
16:27 carandraug pyrimidine, well, there's what to do with Bio::Range and Location, and the Bio-Cluster which I already mentioned
16:27 pyrimidine Bio::Cluster has more dependencies for other modules that vice versa (e.g. little code outside of Bio::Cluster actually uses Bio::Cluster that I have seem)
16:27 carandraug if we gonna remove Bio-Root in the future, why not have it with the Bio-Core. I'd start by making a release of Bio-Core after identifying it.
16:28 pyrimidine *seen
16:28 pyrimidine that seems plausible
16:29 pyrimidine You could simply rename the repo to Bio-Core
16:30 carandraug pyrimidine, but cluster is an interface. Writing a Bio::Cluster::* module requires more knowledge of the object being clustered than of the Bio::Cluster interface. That's why I'd argue to make a Bio::clusterI into core, and Bio::Cluster::unigene into Bio-Unigene, Bio::Cluster::dbsnp into a Bio-variation distribution, and so on
16:31 carandraug pyrimidine, but I want to have a proper plan now. Rewriting the git history will be a pain if we decide to go back later
16:31 pyrimidine carandraug: that's fine
16:31 pyrimidine The interface thing will eventually evolve into something else anyway
16:32 pyrimidine (never liked it frankly.  Too Java-y)
16:32 carandraug Bio::Role :)
16:32 pyrimidine yep
16:32 carandraug pyrimidine, ok. So what goes into Bio-core?
16:33 carandraug I'm guessing to start, all Bio::Root and Bio::*I
16:33 pyrimidine at it's simplest, just the interfaces
16:34 pyrimidine It might be feasible to stick very simple (pure-perl no del) classes there as well (low-level stuff like Bio::Seq), but if splitting everything out works very well I can't see any reason to do so
16:34 pyrimidine Bio::Root could go there
16:34 pyrimidine with the mind that we might migrate away from it
16:35 carandraug pyrimidine, yes.
16:35 pyrimidine have to go to a seminar for about 1.5 hrs
16:35 pyrimidine will back scroll if you have more to add
16:36 carandraug no. I just need to know exactly the way to go. Because it will not be trivial to redo things with a different organization if we change our mind in a month
16:36 carandraug I want this done, but I don't want to rush it too much
17:29 deafferret pyrimidine: genehack ya, only 9 of 129 talks have been accepted http://newsbusters.org/blogs/tim-graham/2​013/03/28/msnbcs-chris-hayes-npr-ick-ive-​watched-very-very-very-little-bill-oreill
17:29 deafferret oops, wrong paste
17:29 deafferret http://www.yapcna.org/yn2013/talks
17:30 deafferret pyrimidine: http://screencast.com/t/m3Ix5onQ
17:35 pyrimidine deafferret: perfect
17:35 pyrimidine there should be a lot of head-smacks
17:43 deafferret woopwoopwoop!
17:43 deafferret is the tag 'perl' redundant at a Perl conference?
17:47 pyrimidine maybe
17:47 pyrimidine but then again, there is now 'perl5' and 'perl6', so....
18:39 denisboyun_ joined #bioperl
19:45 byte48 joined #bioperl
21:19 dnewkirk joined #bioperl
22:30 dnewkirk joined #bioperl
22:49 carandraug joined #bioperl

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