Camelia, the Perl 6 bug

IRC log for #bioperl, 2010-10-04

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

All times shown according to UTC.

Time Nick Message
00:01 carandraug left #bioperl
00:01 carandraug joined #bioperl
00:22 carandraug left #bioperl
01:13 kyanardag_ joined #bioperl
03:27 kyanardag_ left #bioperl
08:06 kai joined #bioperl
08:24 dmb_ joined #bioperl
08:24 dmb_ is now known as Guest69907
08:25 sandyang joined #bioperl
08:27 sandyang left #bioperl
08:42 Guest69907 is now known as dbolser
08:43 dbolser warnings come back to bite me
08:43 kai hehe
08:44 kai that's why they're warnings
08:44 deafferret you know you like it
08:59 dbolser sighk
09:00 dbolser now I feel I shouldn't debug without writing tests...
09:03 daniel__ joined #bioperl
09:04 kai tests++
09:46 dbolser Does the use of Bio::Coordinate:Pair here look correct? http://www.bioperl.org/wiki/​Module:Bio::Coordinate::Pair
09:46 dbolser Example usage
09:54 dbolser hmm... mapping beyond the end of a feature returns the end, mapping before the start returns something else
10:45 dbolser monkey feathers!
10:46 dbolser whats this gap business in bio::coordinate::pair
10:46 dbolser ?
11:10 carandraug joined #bioperl
11:20 dbolser well I found one bug :D
11:20 dbolser not the bug I was looking for but...
11:21 dbolser http://bioperl.pastebin.com/eLf7Dm0f
11:21 dbolser 84 tests before lunch?
11:21 dbolser so... how do I push this script to my fork and get a dev to pull...
11:23 dbolser first thing, bring my remote master in line with origin master?
11:25 dbolser ! [rejected]        master -> master (non-fast forward)
11:37 deafferret git pull?
11:37 deafferret & bed
12:11 * dbolser fears the random
12:12 dbolser git pull
12:12 dbolser Already up-to-date.
12:37 brandi joined #bioperl
12:37 brandi left #bioperl
13:18 dbolser ok, git wins, I give up
13:23 dbolser git seems to respond in two ways, a) everything fine, why would you want me to do anything?! and b) everthing is broken, how can you expect me to do anything?!
13:23 dbolser whats really scarey is when it starts doing things
14:01 ank joined #bioperl
14:07 brandi1 joined #bioperl
14:07 brandi1 left #bioperl
14:08 kai dbolser: hm?
14:09 kai dbolser: did you rebase your local branch?
14:14 tybalt89 left #bioperl
14:39 deafferret .ooOoo__oo... .
14:51 dbolser kai: don't ask
14:51 dbolser mv bioperl-live bioperl-live-git-abortion
14:51 dbolser I found a bunch of bugs in Bio::Coordinate::Pair though :D
14:51 dbolser using -seq_id or not changes the reported coordinates in some circumstances
14:52 dbolser if that ain't bug tastic
15:08 deafferret so you're in your own fork?
15:09 deafferret dbolser: ^
15:09 deafferret git branch -v;  git remote -v
15:15 rbuels joined #bioperl
15:19 dbolser deafferret: what will that do and why?
15:37 Jun joined #bioperl
15:44 pyrimidine joined #bioperl
16:09 kblin hey pyrimidine
16:09 pyrimidine o/
16:09 kblin dbolser: print your available branches and remotes
16:11 dbolser kblin: shant
16:11 dbolser ;-P
16:12 dbolser sorry, I got it sorted
16:12 dbolser pyrimidine: you wrote Bio::Location::Split? (or care for it)?
16:12 dbolser its causing havoc ;-)
16:12 dbolser I think
16:12 pyrimidine I maintain it, but I don't care for it
16:12 pyrimidine :)
16:13 pyrimidine what's causing havoc?
16:13 dbolser I checked in some tests to a branch of my bioperl fork, is it best to point you there or paste the test code?
16:13 pyrimidine paste code and describe the problem
16:14 dbolser http://bioperl.pastebin.com/exbDCXFw
16:14 dbolser something strange is going on with bio::cood::pair when seq-id isn't set
16:15 dbolser I'm seeing warnings from bio::location::split about seq_id's
16:15 dbolser there are two bugs that I think are unrelated
16:15 dbolser the main test errors come around line 300 and on
16:16 dbolser because bio::loc::split fails, the logic of bio::coord::pair fails in these cases
16:16 dbolser and it actually returns different coordinates depending on weather seq_id is set or not
16:19 dbolser part of teh problem is I don't know why bio::coord::pair returns gaps
16:19 dbolser s/problem/debugging challenge/
16:20 dbolser it seems this code can be simplified:
16:20 dbolser http://bioperl.pastebin.com/4kDDczZK
16:22 dbolser could be better: http://bioperl.pastebin.com/nZ5cEiBf
16:23 dbolser splittype3 ?
16:23 dbolser http://bioperl.pastebin.com/6uNwH6nH
16:23 dbolser sorry:
16:23 dbolser http://bioperl.pastebin.com/4smCENHg
16:25 dbolser how many tests test Split.pm?
16:34 dbolser t/SeqFeature/LocationFactory.t
16:35 pyrimidine osrry, stepped away for a second
16:35 pyrimidine *sorry
16:35 dbolser np
16:36 pyrimidine two issues
16:37 pyrimidine one: Bio::Coordinate and Bio::Location functionality has never been fully documented
16:37 dbolser right
16:37 dbolser but I think what I'm seeing are bugs none the less
16:37 pyrimidine two: the current Location system is based on faulty logic (e.g. a split location is supposed to act like a regular simple location)
16:38 * pyrimidine is currently looking at all the pasties
16:38 dbolser sorry, just making a mess of a perfectly good function
16:39 pyrimidine hmm, that could be misconstrued.....
16:39 dbolser the first paste is the test suite showing bugs in bio::coord::pair
16:40 dbolser does this make sense "$value ||= $self->{'_splittype'} || = 'JOIN';" ?
16:40 dbolser one way to find out...
16:43 dbolser my $seqid = $self->seq_id();
16:44 dbolser can I jam in a 'null' seq_id at construction time?
16:45 pyrimidine I'm wondering if this stems from some issues I raised on the list a while back re: sequence ranges and seq_id
16:46 dbolser its very weird... did you run my test set?
16:46 pyrimidine yes, and I'm seeing the same thing
16:46 pyrimidine just so you know, I'm not familiar with Bio::Coordinate (that was Heikki's code)
16:46 dbolser you see that I comment out the seq_id on line 337 and 347
16:47 dbolser try additionally commenting out the seq_id on line 327, and four errors go away
16:47 dbolser (pastebin line numbering)
16:48 pyrimidine yep, can reproduce that
16:48 dbolser freaky eh?
16:48 dbolser not ok 86 <- that one should be tractable
16:48 * dbolser looks
16:50 pyrimidine dbolser: are you seeing the warnings?  "MSG: sorted sublocation array requested but root location doesn't define seq_id (at least one sublocation does!)"
16:50 dbolser yes
16:51 dbolser also (same code, but buried deep in a script): Use of uninitialized value in string eq at /homes/dbolser/perl5/lib/p​erl5/Bio/Location/Split.pm line 190.
16:51 dbolser and Use of uninitialized value in string ne at /homes/dbolser/perl5/lib/p​erl5/Bio/Location/Split.pm line 217.
16:52 dbolser perhasp split is a red herring
16:52 dbolser OK, I'll just change my script to always use a seq_id!
16:53 pyrimidine when comparing sequences, that would be ideal
16:53 pyrimidine otherwise I would consider the location invalid
16:53 dbolser I'm only interested in coordinate conversion
16:54 pyrimidine right, but you can't rely on accurate comparisons w/o having the sequence identity
16:54 pyrimidine chr1:1-1000 is not the same as chr2:1-1000
16:54 pyrimidine but this is one where Bio::Location fails
16:55 pyrimidine (e.g. the sequences are not compared accurately if using something like overlaps)
16:55 dbolser indeed, and I don't expect different coordinates when I give a seq_id or not... given my object usage, nothing is ambigious
16:56 dbolser sorry, read 'this ISN@T where Bio::Location fails'
16:56 dbolser I'm just craming two locations into a pair, one in the in slot one in the out slot
16:58 pyrimidine hah, I like that 'strand' is defined for the Bio::Coordinate::Pair synopsis code, even though the example is a protein sequence
17:04 dbolser I can't keep track of the $match, $value, $results...
17:04 dbolser got to go for my driving lesson
17:05 dbolser pyrimidine: here is the pull request for the tests at least: http://github.com/dbolser/bioperl-live/​tree/dbolser_bio_coordinate_pair_tests
17:08 dbolser http://github.com/bioperl/bioperl-live/pull/2
17:13 randomubuntuguy joined #bioperl
17:15 pyrimidine huh, trying to wrap my head around Bio::Coordinate
17:16 pyrimidine definitely something that needs a good use case
17:18 pyrimidine dbolser: maybe file a bug report?  branches may get lost or go stale.
17:23 pyrimidine dbolser: aaron likes it, I'll pull it into a branch in the main repo, where it can be worked on
17:30 daniel__ Has anyone had this error before?
17:30 daniel__ Replacement list is longer than search list at /usr/share/perl5/Bio/Range.pm line 251.
17:30 pyrimidine that is fixed in bioperl-live
17:30 daniel__ k
17:30 pyrimidine it's a new warning from perl 5.12.x
17:31 daniel__ ahh
17:35 CIA-95 bioperl-live: Dan Bolser topic/dbolser-biocoord_pair_tests * rad8adc9 / t/Coordinate/CoordinateBoundaryTest.t : Adding some more tests for Bio::Coordinate::Pair, showing up a bug - http://bit.ly/dx4B7E
17:35 CIA-95 bioperl-live: Dan Bolser topic/dbolser-biocoord_pair_tests * r4fb0fa0 / t/Coordinate/CoordinateBoundaryTest.t :
17:35 CIA-95 bioperl-live: Found a few more interesting bugs.
17:35 CIA-95 bioperl-live: I'm going to push this before I try to work on a fix... - http://bit.ly/cAh6yd
17:38 pyrimidine dbolser: now in a branch; you should be able to change the integration branch from bioperl:master to bioperl:topic/dbolser-biocoord_pair_tests.
18:35 brandi1 joined #bioperl
18:35 brandi1 left #bioperl
18:51 randomubuntuguy left #bioperl
19:08 kblin oh, that reminds me
19:08 * kblin schedules some time with the hmmer3-parser tomorrow
21:17 carandraug left #bioperl
21:38 pyrimidine left #bioperl
23:37 ank left #bioperl

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