Camelia, the Perl 6 bug

IRC log for #bioperl, 2010-07-21

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

All times shown according to UTC.

Time Nick Message
00:01 carandraug joined #bioperl
00:41 CIA-95 bioperl-live: Duke Leto master * r19d5b75 / t/Ontology/IO/interpro.t : Refactor some tests in t/Ontology/IO/interpro.t - http://bit.ly/cmMBmw
00:59 deafferret Inter PRO!!!
01:00 genehack <sigh> i should have known better than to say I was planning on lurking here -- the Meeting Monsters ate my afternoon.
01:00 genehack but now i'm back, so rbuels, if you're still around, i'll be here for a couple of hours...
01:01 rbuels deafferret: interpro represent.
01:01 rbuels genehack: hi
01:02 * rbuels looks for a bug to point genehack at
01:06 brandi joined #bioperl
01:06 brandi left #bioperl
01:07 * genehack does the happy dance of bug receiving
01:10 rbuels genehack: what part of the pool should I throw you into?  3 feet? 6 feet? 9 feet?  maelstrom?
01:10 genehack let's start around 6-9 feet and see how i do
01:11 rbuels ok ...
01:11 * rbuels looks around
01:15 rbuels lol, most of the bugs still around are either really hairy, or really inconsequential
01:15 * rbuels keeps looking
01:16 genehack well, how about one of each -- i'll use the trivial one to make sure i've got the hang of your contribution flow, and then i'll bang on the hairy one for a bit
01:17 rbuels genehack: how about http://bugzilla.open-bio.org/show_bug.cgi?id=3088
01:17 rbuels this looks doable
01:19 genehack okay, i'll see what i can do...
01:19 rbuels http://lists.open-bio.org/m​ailman/listinfo/bioperl-dev  http://lists.open-bio.org/​mailman/listinfo/bioperl-l
01:19 rbuels genehack: you are probably already on these
01:19 rbuels genehack: well, the dev list you don't really need yet
01:19 rbuels genehack: it doesn't have any traffic yet, and the joins are moderated
01:20 rbuels genehack: so, for your first work, fork the bioperl-live repo on github and then either send a pull request or just ping us here in the channel to take a look at your work
01:20 rbuels genehack: branches should be named topic/somethingorother
01:20 genehack actually, i think i dropped off bioperl-dev a while back -- i'll get back on
01:21 genehack and i've got "hardcore forking action" running as we speak 8^)
01:21 genehack any rule about keeping the bugzilla # in the branch name?
01:21 rbuels genehack: no hard rule, but it's not a bad idea
01:22 genehack ok
01:22 rbuels http://www.bioperl.org/wiki/Using_Git has some rules i guess
01:22 rbuels that page needs some work, since we only recently migrated the main repo to git
01:25 * genehack once again resolves to write that "fork a repo, clone it locally, add the master as the upstream remote" helper...
01:25 rbuels genehack: one non-ideal characteristic of the bioperl codebase that you will probably see very quickly is its gratuitously deep inheritance hierarchies with lots of interfaces, which makes a lot of thinigs difficult
01:26 rbuels genehack: another charming characteristic is the way-way-too-large bioperl-live distro, which makes dependency management a nightmare
01:27 rbuels genehack: and third, test coverage can be quite spotty.
01:27 genehack i remember the "OO is good, so let's do it A LOT!" design from the last time I looked under the covers. 8^)
01:27 rbuels yeah ...
01:28 rbuels actually, an interesting chestnut about bioperl's history:  in the beginning, the thought was to make bioperl and biojava have mirror-image object hierarchies
01:28 rbuels i think that may have been the origin of some of the interface-happiness
01:29 rbuels 'the beginning' here is around 1999-2000
01:29 rbuels bioperl has a lot of history.
01:30 rbuels pyrimidine (aka cjfields) has a couple of embryonic projects that reimplement some of bioperl in a.) Moose and b.) perl6
01:30 genehack so, do people just suffer with the interactive Build.PL, or is there a helper script for "install the common deps" somewhere?
01:30 rbuels (http://github.com/cjfields)
01:30 genehack yeah, i'm watching biome (the Moose one) on github
01:30 rbuels genehack: people just suffer through it
01:31 rbuels genehack: all this optional-deps crap is a symptom of the way-too-large distro
01:31 genehack sure.
01:31 rbuels genehack: if you run the Build.PL with something on stdin, it will take the defaults
01:31 rbuels genehack: which are thought to be fine most of the time
01:32 rbuels 'thought to be'
01:32 * rbuels chuckles
01:32 genehack and it only hurts the first time, right? 8^)
01:32 rbuels well, no, it hurts all the time.
01:32 rbuels lol
01:33 rbuels for the 1.7 release, objective 1 is to break bioperl-live into reasonably-sized distros
01:33 rbuels without optional deps
01:34 genehack sounds like an excellent plan.
01:34 deafferret not gonna fight my bug from 2 days ago? -sigh-
01:34 deafferret it *SO* important!
01:34 deafferret (not really)
01:34 rbuels actually, that could be a good medium bug also
01:34 rbuels deafferret: could you give genehack the bugzilla link?
01:34 rbuels deafferret: or wherever that bug ended up?
01:35 deafferret http://bugzilla.open-bio.org/show_bug.cgi?id=3120
01:36 genehack cool.
01:36 deafferret ice cold
01:36 * genehack boggles
01:37 rbuels hmm...who dat is?
01:37 genehack that may be the deepest module path i've ever seen
01:37 genehack needs a ::ButMaintained at the end for max effect
01:37 rbuels ahhaah
01:38 * deafferret was quoting the Black Eyed Peas
01:39 rbuels well fine then.
01:39 CIA-95 bioperl-live: Duke Leto master * r7314afb / (3 files in 3 dirs): Add Conserved_site as an InterPro root term and update relevant tests - http://bit.ly/9MEKJQ
01:40 deafferret interPRO!!!
01:41 * deafferret is happily converting 10 years of crap out of Exchange into iCal, "Address Book"
01:41 deafferret (sync'd to google)
01:44 rbuels genehack: so which bug is your favorite
01:45 deafferret 3120!
01:45 genehack deafferret's has more potential, i think 8^)
01:45 deafferret \o/
01:46 genehack i think having two will be nice, actually -- it's good to be able to ping-pong between things when you get stuck
01:46 genehack thanks for the "give Build.PL something on STDIN" tip, that was much less painful...
01:49 deafferret umm... gmail Contacts iz confuzing
01:49 deafferret "My Contacts"  134   "All Contacts"  250    -- uhh what?
01:49 deafferret old crap I deleted is still in "All Contacts"
01:50 deafferret like this "Robert Buels" guy
01:50 deafferret all it says is "Corn Hater"
01:50 genehack deafferret: it really wants to keep track of everybody you've ever corresponded with
01:50 rbuels lol
01:50 genehack like, REALLY really.
01:50 rbuels ok guys i'm going to head out
01:51 genehack night
01:51 rbuels genehack: good luck, i'll be around this channel, mostly during business hours pacific
01:51 rbuels dinner &
01:51 deafferret 3120 done already?! woot!
01:51 deafferret :)    gnight
01:51 genehack heh.
01:54 deafferret http://www.youtube.com/watch?v=XuKjBIBBAL8   Cordyceps: attack of the killer fungi - Planet Earth Attenborough BBC wildlife
01:55 * deafferret blames stajich
01:57 deafferret woot! self-healing gmail contacts
01:57 * deafferret submits to assymilation
01:57 deafferret assimilation?
01:59 genehack yeah, that too
02:02 * genehack opens a prophylactic beer and dives into 3120
02:05 brandi joined #bioperl
02:06 brandi left #bioperl
02:27 deafferret holy crap... google stored all my jabber chats from 2008!
02:27 deafferret 2007! dear god
02:27 deafferret what _don't_ they know about me?
02:27 deafferret _dont_
02:27 deafferret huh. irssi underlining weirdness
02:28 brandi joined #bioperl
02:28 brandi left #bioperl
02:30 genehack so, i'm not really up to date on GFF vs GFF3 -- does this look at all close to correct?
02:30 genehack http://paste.scsys.co.uk/46944
02:30 deafferret if those are tab delimited then I would guess yes
02:30 genehack i think they were tab-delimited before i pasted them.
02:31 deafferret that trips me 50% of the time I work with gff  -sigh-
02:31 genehack now, this doesn't actually solve the problem, because i've still got the no_blobs(1) line commented out
02:31 genehack and if i uncomment it, i still get completely empty lines
02:32 genehack (actually, not true -- they've got '.' and 'feature' sprinkled amid a bunch of tabs.)
02:33 genehack but i understand at least part of the problem, which is that (ironically) Bio::SeqFeature::Lite implements a method that Bio::SeqFeature::Generic doesn't.
02:36 deafferret heh   :)
02:38 genehack and now that i'm hip-deep in Bio::SeqFeatureI, i suspect bp_seqfeature_gff3.PLS should be doing things differently. maybe.
02:38 * genehack ponders whether there's enough beer on hand for this
02:41 * genehack throws his hands up
02:41 deafferret :)
02:41 genehack i need an more experienced opinion, because what i'm thinking about doing seems major-ish.
02:42 deafferret KA-BOOM!
02:42 deafferret does that help?   :)
02:42 genehack so, Bio::SeqFeature::Generic nad Bio::SeqFeature::Lite are both supposed to implement Bio::SeqFeatureI
02:43 genehack but B::SF::G::gff_string() and B::SF::L::gff_string() have totally different method signatures -- the one in B::SF::L doesn't match B::SFI; B::SF::G does
02:44 genehack and B::SF::Lite has gff3_string(), which isn't in SeqFeatureI _or_ B::SF::Generic -- instead, according to the interface, if you want to dump something other than GFF2, you're supposed to pass in a formatter object that implements a gff_string() method that outputs what you want
02:45 genehack and, frankly, B::SF::L smells like somebody sort of said "fuck THAT noise"
02:45 deafferret :/   no response to my emails to bioperl-live nor gbrowse
02:45 deafferret so I guess you're free to do whatever without breaking any tests
02:45 deafferret regretting being bored yet?   :)
02:46 genehack so, option #1 is rip gff_string() and gff3_string() out of B::SF::Lite and re-implement them in a BioSeqfeatureI-compliant way
02:47 genehack option #2 is munge bp_seqfeature_gff3.PLS so it introspects the feature it's dumping and does different stuff depending on whether $feature->does('gff3_string')
02:47 genehack option #1 seems likely to break a bunch of stuff and option #2 makes me throw up in my mouth.
02:48 genehack and no, not yet, this is sort of fun.
02:50 genehack BTW, on a totally different topic, what version of Perl is BioPerl targeted against?
02:52 deafferret 5.6.something and up?
02:53 genehack aight, i'll avoid 'say' 8^)
02:53 deafferret ya.
02:55 rbuels genehack: oh christ you have already found the Shame of the BioPerl
02:55 rbuels genehack: which is that FeatureIO and all the feature stuff is a goddamn mess
02:56 rbuels genehack: see here: http://bugzilla.open-bio.org/show_bug.cgi?id=2513
02:57 rbuels http://bioperl.org/wiki/GFF_code_audit
02:57 rbuels see, people have been poking around this for a long time now, but nobody has had enough tuits to really roll up their sleeves and take a machete to it
02:58 * rbuels sits in Bailey's Taproom in Portland, which is a damn fine place to have a beer
02:58 * genehack tiptoes around the minefield
02:59 genehack for this particular problem -- which i hasten to point out, *deafferret* uncovered 8^) -- it seems like the original SeqFeatureI style "pass in a formatter object" would work just fine
02:59 genehack (aside from the whole "change the way SeqFeature::Lite works and break a bunch of stufF" aspects, that is.)
03:04 genehack rbuels: also, looking at 3308 -- and maybe this is just a sign i don't grok the problem -- but it doesn't seem like the bug report includes the BAM file that would be needed to replicate the issue (the one referred to by http://bugzilla.open-bio.org/attac​hment.cgi?id=1509&amp;action=view )
03:15 rbuels joined #bioperl
03:16 * rbuels looks
03:19 rbuels oh that's interesting.  apparently it took me 15 minutes to get your thing about the bam file not being there
03:20 rbuels genehack: yeah i agree.  could you add a comment asking MAJ to provide that?
03:20 rbuels genehack: he's usually pretty responsive
03:21 genehack ok
03:22 rbuels the connectivity in here is rather spotty
03:25 genehack but, on the plus side, i imagine there's beer.
03:26 rbuels there is beer.
03:28 genehack rbuels: would sending something to bioperl-l about the different ways of resolving jay's bug be a productive exercise, or should i just pick one and forge ahead?
03:29 rbuels genehack: if you did that, you would be opening a can of worms too big for mailing list discussion, i think
03:29 genehack okay, that's sort of what i was figuring.
03:30 rbuels genehack: probably the best course of action would be to fix the bug with minimal disruption, and *then* comment on the mailing list about that can of worms
03:31 genehack okay, that argues for the "make bp_seqfeature_gff3.PLS introspect the feature and do different stuff" approach, i think.
03:31 rbuels genehack: because there will be no agreement on a course of action for the bug in question.  but the post-comment might at least serve to catalyze some discussion about the shortcomings of the current situation.
03:31 genehack i'm also now wondering if i haven't gotten side-tracked -- because the way he _wants_ to do it, without blob storage, is failing in a different way and i think for a different reason.
03:32 rbuels the way who wants to do it?
03:33 genehack ah, sorry -- the test case in the bug has a commented out '$db->no_blobs(1);' call
03:33 * rbuels looks
03:33 deafferret indeedly
03:34 genehack if you uncomment that, the failure is totally different -- it looks like the feature that gets read back in by bp_seqfeature_gff3.PLS isn't getting instantiated properly, or the script is trying to use the wrong interface for it, or something of that nature.
03:35 genehack so i'm now thinking that maybe i should attack that aspect first, because maybe that fixes the bug deafferret cares about and skirts the FeatureIO mess
03:36 * deafferret does the mashed potato
03:37 genehack and on that note, it's bedtime:30 here, so i'll see y'all tomorrow...
03:37 deafferret o7
03:37 rbuels genehack: alrighty.  thanks for the work so far.
03:39 rbuels yeah, i think genehack is right about failing in a different way and for a different reason when no_blobs is on versus off.
03:41 rbuels and, without delving into it too deeply (not really enough tuits), the approach of special-casing based on feature type in the dump script sounds reasonable.
03:54 deafferret pretty sure my bug report shows with and without no_blobs(1)
03:54 deafferret yes, totally different fail (on my boxen)
04:07 rbuels that was silly of me to rebase the leto interpro branch
04:07 rbuels makes things confusing.
04:08 * rbuels merrily slaps himself on the wrist
04:10 deafferret interPRO!!!
04:10 rbuels fo sho.
04:10 deafferret PROfessional? or PROstate?
04:11 rbuels PROlapse
04:11 deafferret PROlonged suffering?
04:27 dukeleto mmmmm, suffering
04:30 rbuels dukeleto: wow.  half of oscon is converging on bailey's.
04:30 rbuels dukeleto: mst is here, and he said bunches of other perl people are converging here momentarily
04:31 deafferret hasn't mst's visitor visa expired yet?
04:31 deafferret GO HOME WHITE HAIR
04:31 rbuels hehe
04:31 rbuels yes he does have white hair right now
04:32 dukeleto rbuels: i am back at backspace with my buddies
04:32 dukeleto rbuels: i could stop over there, but it is pretty chill here
04:34 * rbuels nods
04:35 rbuels dukeleto: duckyd says you should at least make a rebastiality tweet
04:44 deafferret it is also pretty chill in the deafferret cube
04:44 deafferret but now, bed
04:45 rbuels deafferret: nighty night
06:06 bag joined #bioperl
06:34 bag joined #bioperl
07:43 * dukeleto just got home
08:36 dukeleto joined #bioperl
09:45 vinnana joined #bioperl
10:34 brandi joined #bioperl
10:34 brandi left #bioperl
12:40 brandi joined #bioperl
12:40 brandi left #bioperl
13:36 brandi joined #bioperl
13:36 brandi left #bioperl
14:00 brandi joined #bioperl
14:00 brandi left #bioperl
14:24 brandi joined #bioperl
14:25 brandi left #bioperl
15:06 brandi joined #bioperl
15:06 brandi left #bioperl
16:13 deafferret ...o.
16:17 * genehack has been in meetings since 9am (zone local == 1216)
16:17 * genehack goes off in search of food and some semblance of sanity
16:18 deafferret \m/   meetings!!   \m/
16:40 brandi joined #bioperl
16:40 brandi left #bioperl
17:03 dukeleto joined #bioperl
17:05 brandi joined #bioperl
17:16 ank joined #bioperl
17:53 CIA-95 bioperl-live: Duke Leto topic/leto_interpro_get_dbxref * r7c6739b / t/Ontology/IO/interpro.t : Add a test to show that there is a bug with get_dbxrefs on InterPro terms - http://bit.ly/d9fnMy
18:18 brandi joined #bioperl
18:18 brandi left #bioperl
18:39 pyrimidine joined #bioperl
18:50 brandi joined #bioperl
18:50 brandi left #bioperl
19:00 pyrimidine tornados!  fun!
19:01 deafferret eek!
19:04 * pyrimidine heads to the basement
19:04 * pyrimidine and Toto too!
19:08 * rbuels wonders if pyrimidine will wake up to find that his house has landed on James Gosling, and then be serenaded by a bunch of java midgets
19:12 pyrimidine java isn't THAT evil
19:12 pyrimidine close, though ;-)
19:36 kyanardag_ joined #bioperl
20:01 risha joined #bioperl
21:17 brandi joined #bioperl
21:17 brandi left #bioperl
22:05 pyrimidine left #bioperl
22:37 bag joined #bioperl
22:41 brandi joined #bioperl
22:41 brandi left #bioperl

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