Camelia, the Perl 6 bug

IRC log for #bioperl, 2009-08-14

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

All times shown according to UTC.

Time Nick Message
01:37 ilbot2 joined #bioperl
01:37 Topic for #bioperlis now http://www.bioperl.org - You can get help here, but you must be patient.  Paste to http://gist.github.com/.
03:02 tarbo2_ joined #bioperl
15:22 _tristan_ joined #bioperl
15:23 _tristan_ hello! I just updated my bioperl-run through svn, and it looks like the folder structure has changed.
15:23 _tristan_ We have now bioperl-run/lib/Bio/Tools/
15:24 _tristan_ while before it was just bioperl-run/Bio/Tools
15:24 _tristan_ Right?
15:24 _tristan_ (i.e. the lib/ folder was added)
15:25 _tristan_ in my $PEL5LIB I have /opt/src/bioperl-live/ and /opt/src/bioperl-run/
15:26 _tristan_ now it looks like I should have /opt/src/bioperl-run/lib/
15:28 _tristan_ (s/PEL5LIB/PERL5LIB)
15:38 rbuels joined #bioperl
15:38 TristLef__ joined #bioperl
15:40 TristLef__ (not sure why but I've been disconnected, and reconnected with another nickname, I am _tristan_)
16:02 _tristan_ joined #bioperl
16:20 _tristan_ well, I will post this to the mailing list, since things look very quiet up here
16:24 _tristan___ joined #bioperl
16:26 rbuels _tristan___: yes you're right it has moved
16:37 * deafferret blames rbuels
16:37 rbuels probably fair
16:38 deafferret ya? did you do that?
16:39 rbuels no it was pyrimidine
16:39 deafferret and the truth shall set you free
16:39 deafferret s/^/do try not to fall on your sword needlessly next time. //
16:41 deafferret where did pyr* get all these "be more like the rest of CPAN" ideas like lib/  ??? change sucks
16:41 rbuels lol
16:41 rbuels he's always had them
16:41 rbuels i think i just provided a little more .... impetus?
16:42 rbuels oh oh
16:42 rbuels i think i just provided a little more .... shall we say .... impetus?
16:42 rbuels (the "shall we say" is key)
16:50 deafferret I need to write an rbuels library... http://www.cs.utexas.edu/~jbc/home/chef.html
16:50 deafferret ooo, there it is. this one  http://www.rinkworks.com/dialect/
16:51 deafferret i reckon i jest provided a li'l mo'e impetus
16:51 deafferret bioperl-bot: be rbuels
16:52 * deafferret sighs
16:52 rbuels sgn has a bot.
16:52 rbuels i could put her in here
16:52 rbuels she's a flooterbuck
16:52 deafferret neato.  :)
16:56 rbuels i think she only knows how to connect to one server at a time
16:56 rbuels and i don't want to run multiple....
16:56 rbuels bleh
16:56 rbuels maybe i should do some USEFUL work.
16:57 deafferret do try not to let me down next time
16:57 * rbuels sniffs
16:57 deafferret to not
16:57 deafferret do ... to not
17:03 _tristan_ so this is not a bug?
17:06 deafferret _tristan_: no. It's a feature. The repository layout changed to be more like the rest of CPAN.
17:07 _tristan__ joined #bioperl
17:07 _tristan__ ok, thanks
17:15 rbuels joined #bioperl
19:31 pyrimidine joined #bioperl
19:32 pyrimidine o/
19:33 deafferret o7
19:33 rbuels lol
19:33 rbuels (that's a picture of a little guy with both his hands up the air)
19:33 rbuels (or is it laugh out loud?)
19:33 pyrimidine both
19:33 rbuels (it's ambiguous)
19:34 pyrimidine sendu made the change to be CPAN-like
19:34 pyrimidine (I just needled him into doing it)
19:34 * rbuels likes needling
19:35 rbuels pyrimidine: lots of work on biome eh
19:35 rbuels that bot keeps popping in here
19:35 rbuels ;-)
19:35 pyrimidine yes, it's the automated posting via github.  We need dalek
19:36 pyrimidine (or maybe bioperl-bot could be borged by dalek)
19:40 pyrimidine joined #bioperl
19:42 pyrimidine deafferret: know if #moose is publicly logged (ala #bioperl)?
19:42 Guest95942 joined #bioperl
19:43 deafferret pyrimidine: It's not. I have the logs though, if you want them
19:44 rbuels no no don't look at the #moose logs
19:44 rbuels lol
19:44 deafferret especially not at rbuels purl factoid
19:45 pyrimidine I might.  Trying to work out an issue with aliasing attributes (but I think the latest Moose can do this)
19:45 deafferret k, well, let me know if you want them. I can email them to you if you give me an address
19:46 pyrimidine just msg'd you the email
19:47 deafferret sent
19:48 pyrimidine got 'em.  thx!
19:48 deafferret natch
20:39 rbuels might have a line on a good student for next year's gsoc.  one of the SGN interns this summer is really good.
20:44 pyrimidine rbuels: make sure they submit an idea close to the deadline next year if they are interested; gsoc proposal submission process can be brutal
20:45 rbuels what was the date for this year's
20:45 * rbuels looks
20:45 * pyrimidine looking
20:46 rbuels march 13?  "Mentoring organization application deadline." ?
20:46 rbuels http://socghop.appspot.com/document/s​how/program/google/gsoc2009/timeline
20:46 pyrimidine yep
20:46 rbuels oh, april 3, student app deadline
20:47 rbuels wow that's pretty early
20:47 pyrimidine yes
20:47 pyrimidine so the mentors and students have some lead time to prep prior to coding
20:48 pyrimidine app process can be fairly brutal; I pummeled two bioperl-based proposals for The Perl Foundation b/c they were overly simplistic
20:48 pyrimidine NESCent was easier, but not by much
20:51 pyrimidine wow, that was weird.  Carl Woese just passed by.
20:57 deafferret was his rolling around in the Stanley O. Ikenberry Chair?
20:57 rbuels hahahhaa
20:59 pyrimidine no, just walked by, mr. wikipedia
21:01 deafferret If I had that chair I'd make postdocs carry me around on it on poles
21:01 rbuels well, rather, his bearers walked by.
21:01 rbuels yes
21:01 pyrimidine yes.
21:01 deafferret the original "passed by" certainly allowed for lots of creative imagery
21:01 * rbuels gives deafferret 2 demerits for making the same joke as him
21:02 deafferret I made it first  :p
21:02 deafferret d:
21:02 * rbuels gives him 2 more demerits for impersonating doy
21:03 * deafferret rips off an rbuels leg, eats it like a Ren Fair
21:03 rbuels you flubbed it.  like AT a Ren Fair
21:04 pyrimidine don't know if you can see it, but the odd statues in front of our building are supposed to be ribosomes:
21:04 rbuels or more correctly "as if he were at a Renaissance Fair"
21:04 pyrimidine http://128.174.122.42/home/homeJ.html
21:04 pyrimidine he supposedly hates 'em
21:04 rbuels the sony network camera does not like me
21:04 pyrimidine (they look like giant Nerds)
21:04 rbuels lol
21:05 rbuels i fail to see how you could *ever* make a statue of a ribosome that would be at all attractive
21:06 pyrimidine yes, and added to that is the 'artistic license'; all the ribosomes are exactly the same, just different sizes and colors
21:07 deafferret pyrimidine: go outside and wave to us  :)
21:07 pyrimidine ok
21:07 rbuels oh wow it's a slow friday afternoon
21:07 rbuels lol
21:07 deafferret then have Carl's posse go by
21:07 rbuels with the enclosed sedan chair
21:08 deafferret popemobile
21:08 rbuels omg i see him
21:08 rbuels he's wearing a blue shirt
21:08 deafferret is that him? did he wave?
21:08 rbuels yeah
21:08 rbuels he's standing by one of the statues now
21:08 rbuels hah
21:09 * deafferret laughs
21:09 rbuels for some reason the cam works in konq, but not firefox
21:09 deafferret ya. I'm in IE
21:09 deafferret pyrimidine++
21:09 rbuels that was amusing
21:09 pyrimidine back
21:09 deafferret good on ya  :)
21:10 rbuels you were just a little speck
21:10 rbuels but i could tell, you were the only one not purposefully walking anywhere
21:10 deafferret You guys watch -- I'll dynamite one of the cranes here  http://www.unmc.edu/blog/publicaffairs/?cat=12/
21:10 pyrimidine yeah, they have it zoomed all the way out
21:10 rbuels hahahaha
21:12 deafferret can't find a webcam here. bummer. oh well, back to this documentation slog
21:45 pyrimidine opened two core branches for playing, focused on refactoring Align/LocatableSeq and GFF/Features
21:47 rbuels hmmmmm
21:48 deafferret you should get your own chair once that's complete
21:48 deafferret must get tired of standing
21:48 rbuels make sure it has casters.
21:48 pyrimidine I intend on cleaning up that shit
21:48 deafferret casterators?
21:48 * rbuels svn switches to TRY_gff_refactor
21:49 pyrimidine feel free to tear into FeatureIO and anything GFF-related there.  It's in dire need of cleaning up.
21:49 * rbuels goes to gff refactor wiki page
21:51 rbuels so bsfa is a typedseqfeaturei now...
21:52 pyrimidine don't feel limited to that page.  and don't feel obligated to keep bsfa.
21:52 pyrimidine it's too 'heavy' (it converts literally everything into Annotation)
21:52 rbuels yeah it really is heavy as hell
21:52 * rbuels remembers going through it a couple years ago
21:55 pyrimidine This will be easy to do with Moose, as we can have a simple attribute with the ontology term attached ('type'), use a predicate (is_typed), etc.  Use a factory to add the term in if possible, warn otherwise.
21:57 rbuels i vaguely remember the factory idea...
21:57 rbuels you'd have a featurefactory right....
21:58 rbuels (which would be used internally by featureios)
21:58 pyrimidine No, but you can use the generic object factory
21:58 rbuels make a gist with some quasi-code illustrating what you're talking about
21:58 pyrimidine though it may be faster to use direct_from_new(0 instead of new() (hacky, but faster)
21:59 rbuels so what we need is a lighter-weight typed feature object
21:59 pyrimidine yes
21:59 rbuels but what's the factory thing?
22:00 pyrimidine it'll hold all the necessary things for typing the features
22:00 rbuels the ontology object....
22:00 rbuels attributes for whether to be making the features typed
22:00 pyrimidine Ontology-related, unflattening, etc
22:00 pyrimidine yes
22:01 pyrimidine basically, just abstract that out
22:03 pyrimidine I suppose FeatureIO itself could act as the factory; set the bells and whistles you want, then call next_feature to iterate through the sf's (which would be typed, unflattened, affixed w/ unique ID, whatever)
22:03 rbuels add a boolean typed() switch to the featureIO base
22:03 pyrimidine yes
22:03 rbuels and....
22:04 rbuels make all the subclasses use some FeatureIO base method to construct features
22:04 rbuels i like that solution better
22:04 pyrimidine yes, exaactly
22:04 pyrimidine s/exaactly/exaaaaactly/
22:04 rbuels so how to avoid the serialization problem
22:04 rbuels cause features will have to have a reference to the ont somewhere
22:05 * rbuels perldocs Store
22:05 * rbuels perldocs Storable
22:05 pyrimidine That's the tricky bit, yes, but the overhead is too high to store the entire ontology with each feature (which is what appears to be happening)
22:06 rbuels each feature doesn't actually have a whole copy of the ont in it
22:06 rbuels they just all have refs to the same ont object
22:06 rbuels and storable is dumbly following it
22:06 rbuels (is my theory)
22:06 pyrimidine yes
22:07 pyrimidine that's pretty much it
22:07 rbuels maybe the objects don't have to have a ref to the ont
22:07 rbuels feature objects i mean
22:07 pyrimidine The problem is the ontology is loaded in as a class attribute
22:08 rbuels "
22:08 rbuels Any class may define hooks that will be called during the serialization and deserialization process on objects that are instances
22:08 rbuels of that class.
22:08 rbuels (in storable docs)
22:09 pyrimidine so that's an alternative (use a hook to clear the ontology prior to serialization)
22:10 rbuels i think that's better, because you want to be able to find the ontology if you are given some feature or term
22:10 rbuels even if the feature just has a reference to the term, the problem is the same, since storable will crawl through to the ont through the term
22:10 rbuels gotta have these links.
22:11 rbuels just have to be careful with serialization
22:11 * pyrimidine wondering if the ontology can be loaded lazily if it isn't present
22:12 rbuels oh sure.
22:12 rbuels probably the features will just have references to terms.
22:13 rbuels and the terms will have references to ontologies
22:13 rbuels and the terms will just be serialized as strings : 'SO:123423'
22:13 rbuels and the string will be inflated when accessed
22:13 pyrimidine yes
22:13 pyrimidine makes sense to me
22:14 rbuels you mentioned some cruft in featureIO itself?
22:14 rbuels or featureIO::*?
22:15 pyrimidine yes, but I think much of it's related to the same overall problem
22:15 pyrimidine whether or not to enforce typing, etc
22:15 rbuels gff.pm uses bsfa's right now
22:15 pyrimidine but I think other's don't
22:15 pyrimidine there's inconsistency iirc
22:15 rbuels ok, so doing this schema and porting all the parsers to it should do the trick
22:15 rbuels s/schema/scheme/
22:16 rbuels i've got a new parser to put in there actually
22:16 rbuels for genomethreader xml
22:16 rbuels i'll put that in there while i'm at it and port it over too
22:16 pyrimidine I think so.  It's much simpler, it'll just piss off anyone using the old way of doing things (via bsfa)
22:16 rbuels whatever.
22:16 pyrimidine sounds good
22:18 rbuels whaa....
22:18 rbuels ack-grep FeatureIO shows only FeatureIO/*.pm
22:18 * rbuels smells a splitoff
22:19 rbuels and FeatureIO.pm of course
22:19 rbuels but that will be after the refactor
22:19 pyrimidine yes
22:19 pyrimidine Another thing to look into (before I forget) is whether to add the option to normalize everything to AnnotationI
22:19 rbuels ?
22:19 pyrimidine but that can be added in as an option
22:20 * rbuels looks at annotationI
22:20 pyrimidine sfs can hold simple tags and AnnotationI (sfs are AnnotatableI).  This is a bit inconsistent.
22:22 rbuels simple tags = strings, right?
22:22 pyrimidine bsfa hard-coded everything to be AnnotationI (no choice).  tags/values = simple strings
22:22 pyrimidine yes
22:22 rbuels i'm not a big fan of annotationi
22:22 pyrimidine Neither am I
22:22 rbuels at least the method names...
22:23 rbuels so...there is a need to have something more than simple strings i guess
22:24 pyrimidine Well, we could come up with something like a lazy annotation collection
22:24 rbuels hmmmm
22:25 pyrimidine store everything as tag/value strings (or groups of tag/values), and create them on the fly
22:26 rbuels there are some mailing list threads about the need for object annotations aren't there?
22:26 * rbuels looks
22:27 deafferret are you using moose and doing all this in the biome github?
22:28 rbuels nono, this is in a branch of regular bioperl
22:28 pyrimidine no, this is in core, but will be ported over
22:28 deafferret and no moose?
22:28 rbuels bioperl-live/branches/TRY_gff_refactor
22:28 rbuels no moose.
22:28 pyrimidine at least not yet  ;)
22:28 rbuels one featureIO is split off, moose could be added as a dependency
22:29 rbuels it's going to be the first thing i split off
22:29 pyrimidine actually, that's not a bad idea
22:29 rbuels but not yet.
22:29 rbuels well.
22:30 pyrimidine you could probably use the branch for that purpose
22:30 pyrimidine it's fairly isolated
22:30 rbuels quite isolated.
22:30 rbuels ack-grep FeatureIO
22:30 rbuels only returns FeatureIO*
22:31 pyrimidine the only thing that might be required is to mess with the SeqFeatureI interface if you want to add some additional methods
22:31 pyrimidine is_typed(), term(), whatever
22:31 pyrimidine but that's simple
22:32 rbuels well the split is a very simple one
22:32 pyrimidine yup
22:32 rbuels well.
22:32 rbuels hmmm i can't decide.
22:33 pyrimidine I know what I've decided, and that's a beer
22:33 pyrimidine or two
22:33 rbuels hahahaha
22:33 rbuels well said.
22:33 deafferret central time zone: dismissed. west coast: back to work!
22:33 * rbuels writes an implementation plan on the GFF_refactor page
22:34 pyrimidine you can work on the branch and split off sooner or later.  It won't matter until the point of a merge.
22:34 pyrimidine and then everyone will be pissed off  ;)
22:34 rbuels hooray!
22:34 * rbuels is a specialist at that
22:35 deafferret "merge" meaning FeatureIO:: goes poof! into its own universe?
22:35 rbuels at some point
22:35 rbuels it's a really easy one.
22:35 rbuels grep -Rl FeatureIO Bio/
22:35 rbuels nothing else uses it
22:37 * pyrimidine was on phone
22:37 pyrimidine yes that's essentially it
22:39 pyrimidine It would be a different story if you dragged in Bio::SeqFeature along with it, but that's unnecessary
22:40 rbuels yep.
22:40 pyrimidine Ok, heading out for a beer.  Back on monday!
22:40 rbuels ok, ttyl
22:41 deafferret o7
22:41 * rbuels puts in work, represents the set
22:42 * deafferret goes loco with the cake
22:42 rbuels omfg
22:42 rbuels i forgot about that
22:42 pyrimidine left #bioperl
22:43 deafferret I still don't know what it means, but I'm pretty sure it make my ass badder.
22:43 deafferret ass bladder? what?
22:43 rbuels oh it definitely will do that one way or another
22:57 ptl_ joined #bioperl
23:01 deafferret k. I wrapped up that mess for now. I think I'm going to commute before it gets dark (motorcycle)
23:04 rbuels deafferret: bon voyage
23:06 deafferret g'night. don't run out of cake
23:06 deafferret infinite loop:  http://www.youtube.com/watch?v=yv3Qm69pocA
23:10 rbuels oh yeah
23:11 rbuels this time i have my sub hooked up

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