Camelia, the Perl 6 bug

IRC log for #bioperl, 2011-04-07

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

All times shown according to UTC.

Time Nick Message
00:17 rdesfo left #bioperl
02:01 dukeleto left #bioperl
02:01 kai left #bioperl
02:02 randomubuntuguy left #bioperl
02:03 randomubuntuguy joined #bioperl
02:03 dukeleto joined #bioperl
02:03 kai joined #bioperl
02:27 randomubuntuguy left #bioperl
02:34 CIA-119 bioperl6: takadonet master * r34a45e4 / lib/Bio/Annotation/TypeManager.pm : implemented is_valid method in TypeManager. No test for it - http://bit.ly/h383v0
03:06 CIA-119 bioperl6: takadonet master * r7356d29 / t/SeqIO/fasta.t : cleaned up test file alittle - http://bit.ly/ehddgE
03:19 CIA-119 bioperl6: takadonet master * rf46a4ba / (t/SeqIO/genbank.t t/data/roa1.genbank): added test genbank file. Imported genbank.t file from bioperl-live. Change most to Perl 6 but still all mostly commented since SeqIO::Genbank is NYI. - http://bit.ly/dKiBYd
03:44 Drin left #bioperl
03:45 Drin joined #bioperl
03:54 zenman_ left #bioperl
05:58 mmlevitt joined #bioperl
05:59 ank left #bioperl
06:01 ank joined #bioperl
06:11 mmlevitt left #bioperl
07:43 melic_ is now known as melic
12:40 reecehart joined #bioperl
13:26 reecehart left #bioperl
14:30 * dbolser 's impossible question of the day...
14:31 dbolser can I use the Bio::DB::SeqFeature::Store's dsn to pass 'per connection' variables such as max_allowed_packet=32M?
14:31 dbolser trying:        -dsn     => 'www-potato:mysql.compbio.dundee​.ac.uk;max_allowed_packet=32M',
14:31 dbolser but no joy
14:31 dbolser (apparently)
14:39 dbolser crud...-dbi_options      A hashref to pass to DBI->connect's 4th argument, the "attributes."
14:39 dbolser hmm
14:44 pyrimidine dbolser: not sure, but I don't think so.  You can probably grab the DBI-based handle from the store and set those there (maybe)
14:45 dbolser pyrimidine: looks like setting my dsn to inclue the setting should work, but apparently doesn't
14:45 dbolser http://search.cpan.org/~capttofu/​DBD-mysql-4.018/lib/DBD/mysql.pm
14:46 pyrimidine I don't think the 'dsn' for DB::SF::Store is the same as the 'dsn' for DBI/DBD::mysql
14:46 pyrimidine or maybe it is
14:46 dbolser looks so (roughly)
14:47 dbolser $dsn = "dbi:mysql:$dsn" unless $dsn =~ /^dbi:/;
14:47 dbolser $dbh = DBI->connect($dsn,$user,$pass,$dbi_options) or $self->throw($DBI::errstr);
14:47 dbolser line 218 of sfs:mysql.pm
14:48 pyrimidine yeah, but I don't think addtional args beyond user/pass and the raw dsn are passed along to DBI
14:48 dbolser pyrimidine: the args are part of the dsn string
14:48 dbolser and it looks like they are there when the dbh is created
14:48 pyrimidine actually, looks like you can pass the options in as well
14:49 dbolser I'll grab the dbh and see if I can probe it
14:49 pyrimidine -dbi_options
14:49 dbolser didn't work either
14:49 pyrimidine not sure why it wouldn't
14:49 dbolser perhaps there is a server side limit on what I can set re: max_allowed_packet
14:50 pyrimidine possibly
14:50 dbolser I tried setting it really small, but that didn't seem to affect anything
14:51 pyrimidine could also be the way the wrapper around the dbh (SF::Store::DBI::mysql) is designed, e.g. the way it is called
14:51 dbolser looks like it gets passet right through
14:51 pyrimidine you can check the dbh directly by calling $store->dbh
14:51 dbolser but print Dumper $store->dbh; # gives me $VAR1 = bless( {}, 'DBI::db' );
14:52 dbolser (after connecting)
14:52 pyrimidine yes, that tends to happen with XS-based code
14:52 dbolser ah...
14:52 * dbolser searches for a 'give me all your settings' option
14:53 * dbolser tries mysql_read_default_file= as a sanity test
14:54 dbolser yup, it is picking that up
14:54 dbolser yup, it is picking that up    -dsn        => "$DSN;mysql_read_default_fi​le=/homes/dbolser/.my.cnf",
14:54 dbolser If I criple that file, the perl script fails
14:54 pyrimidine look under 'Database Handles' for supported attributes
14:55 dbolser ty
14:55 pyrimidine http://search.cpan.org/perldo​c?DBD::mysql#DATABASE_HANDLES
14:56 dbolser yup
14:57 pyrimidine looks as if 'max_allowed_packet' is not supported
14:57 dbolser I've now set it in my.cnf, which I know is being picked up
14:58 pyrimidine probably the safest way
14:58 dbolser no joys
14:58 dbolser how do dump connection variables?
14:58 dbolser from $dbh
14:59 dbolser I dumped all these, but nothing http://search.cpan.org/~capttofu/DBD-mysq​l-4.018/lib/DBD/mysql.pm#DATABASE_HANDLES
15:04 dbolser show variables; # /me should be able to find it
15:07 dbolser seems I can't vary it from the client side
15:09 xinwei joined #bioperl
15:13 dbolser wow... mysql was just being a git when it came to notation
15:14 dbolser oh no... dba has bumped it for me, my settings still aren't having an effect
15:18 xinwei left #bioperl
15:19 dbolser here is my new proposed bp_script:
15:19 dbolser https://gist.github.com/907984
15:19 dbolser bp_seq_load.pl edit
15:31 pyrimidine dbolser: is that something you would like to add to bp-live?
15:39 dbolser pyrimidine: I guess... you want me to write tests for it?
15:40 dbolser I'm still only assuming it works at this stage, because I accidentally wiped my feature database when trying to add sequences
15:40 dbolser I'm hoping this will let me return the sequence of transcripts given genomic DNA and 'gene' GFF?
15:40 dbolser (spliced)
15:41 dbolser given I've loaded my landmarks, can I load sequences in parallel to features?
15:53 rdesfo joined #bioperl
15:56 pyrimidine rdesfo: o/
15:56 rdesfo pyrimidine:hello
15:59 pyrimidine rdesfo: going back over your proposal to add some comments
16:03 rdesfo I have "perl6 now", "perl programmingfor medicine and biology", "modern perl" as  well as "Perl and Apache" and "Mastering Algorithms with perl"
16:09 pyrimidine rdesfo: here is the in-progress book: https://github.com/perl6/book
16:09 pyrimidine rdesfo: PDF here: https://github.com/downloads/​perl6/book/2011.01.letter.pdf
16:10 pyrimidine it's not complete, but there are plans afoot to add more to it for this summer's YAPC:NA
16:10 rdesfo thanks... I'm have that as well and I'm also watching it on Github
16:10 rdesfo Nice
16:12 pyrimidine so, there are a number of things that can go into a project such as this, but I think it's a good idea to establish up-front why this project has to be done in perl  vs python/ruby/perl5/etc (the power of grammars)
16:12 pyrimidine *perl6 vs
16:13 pyrimidine specific code examples of a simple grammar might demonstrate this
16:14 rdesfo ok
16:15 pyrimidine the JSON example in the book is nice, but you could also use the FASTA grammar that takadonet added to bioperl6
16:24 pyrimidine probably should also tie in a simple Action class to that grammar (also in bioperl6)
16:24 rdesfo ok
16:25 * pyrimidine wonders if there is a nicer way to display code examples on Melange....
16:29 pyrimidine rdesfo: aha! http://www.google.com/support/forum/p/Google%​20Docs/thread?tid=715436a20b928333&hl=en
16:29 pyrimidine the code highlighter supposedly supports perl 6
16:29 pyrimidine that might be interesting....
16:30 rdesfo ok
16:31 pyrimidine rdesfo: did you understand what I meant by tying the tokens from various formats to similar rules?
16:31 pyrimidine s/rules/actions/
16:34 rdesfo I know that Regex in Perl 6 is broken up into tokens and rules
16:34 pyrimidine right
16:36 pyrimidine rdesfo: the trick is, what do you actually do with the data from the match object generated from the grammar?
16:36 pyrimidine in general, you can attach a set of actions to each token/rule
16:36 rdesfo ok
16:37 pyrimidine in the 'Using Perl 6' book, it's in Chapter 10 (specifically, 10.2 'Extracting Data')
16:38 pyrimidine if the tokens/rules are named similarly or have a common parent Grammar, one can tie similar actions to each of those tokens
16:38 pyrimidine one set of actions could be to create an object
16:39 pyrimidine another could be to load a database with the data
16:39 pyrimidine format the data into a JSON stream
16:39 pyrimidine and so on
16:40 rdesfo I'll make sure I go over that again
17:00 rdesfo how do you feel about the proposal now?  Is there any other updates that I should make?
17:04 rdesfo I'm sorry, but I have to go... I have an appointment at 2.  I'll check the document when I get home for any other comments
17:04 rdesfo thank you very much for the help today
17:05 rdesfo left #bioperl
17:27 dbolser hmm
17:28 dbolser so I loaded sequences into the seqfeature store using the above script, but now my features don't have sequences attached
17:28 dbolser should I first have queried the landmark feature, attached the sequence and shoved it back?
17:29 pyrimidine dbolser: are these SF::Generic?
17:29 rbuels dbolser: i dunno, look at how bp_seqfeature_load.pl does it.  it loads seqs.
17:29 dbolser rbuels: ohrly?
17:29 rbuels dbolser: sure.
17:29 dbolser I didn't know it did
17:29 * rbuels slaps his forehead
17:29 * rbuels wrings his hands
17:29 pyrimidine dbolser: yep.  loads FASTA
17:29 * dbolser slaps his forhead
17:29 * rbuels cries "mama mia!"
17:29 * dbolser cries
17:29 pyrimidine also embedded FASTA in gff3 files
17:30 dbolser pyrimidine: right, I got hung up on that last mode
17:30 pyrimidine itza very spiiiicy meatball
17:30 dbolser SF::Generic?
17:30 dbolser so ... since my features and sequences are apparently in the db... can I tie them together with some magic?
17:30 pyrimidine attach_seq()
17:30 pyrimidine but...
17:31 dbolser indeed
17:31 pyrimidine I don't think that will work as expected
17:31 dbolser should I just keep cramming things in?
17:31 pyrimidine If you are passing something through Bio::SeqIO, the default SeqFeatureI class is Bio::SeqFeature::Generic
17:31 pyrimidine not Bio::DB::SeqFeature
17:31 dbolser ahhh
17:32 pyrimidine so, I'm sure that is part of the problem
17:32 dbolser lets see if bp_seqfeature_load.pl fixes it...
17:33 dbolser where is the FASTA setting for bp_seqfeature_load.pl ?
17:34 dbolser I just pass a fasta as a gff?
17:34 pyrimidine yep
17:34 rbuels dbolser: pretty sure you just pass the fasta unmodified
17:34 * rbuels hasn't used it in a few weeks
17:34 dbolser not documented here..., but I'll try
17:34 rbuels documentation?!?!  i heard of that once.
17:34 pyrimidine the GFF loader looks for the FASTA header
17:35 dbolser ic
17:35 dbolser if I submit a doc patch, do I have to write tests?
17:35 pyrimidine thousands of them
17:35 pyrimidine no
17:35 * dbolser swoons
17:36 dbolser loading Assembly/PGSC0003DM/PGSC0003DMB.fa...
17:36 dbolser should I do a famous pull request?
17:36 pyrimidine sure
17:38 dbolser one day I'll learn git
17:41 dukeleto dbolser: or git will learn you
17:43 dbolser true that
17:44 dbolser hrm... --noalias-target is described in the usage, but not part of getopt long
17:47 pyrimidine you mean, there is something undocumented in the script?!?
17:47 * pyrimidine is shocked and awed
17:48 dbolser the opposite
17:52 dbolser this is new...
17:52 dbolser Storable binary image v60.156 more recent than I am (v2.8) at blib/lib/Storable.pm (autosplit into blib/lib/auto/Storable/thaw.al) line 421, at /homes/dbolser/perl5/lib/perl​5/Bio/DB/SeqFeature/Store.pm line 2553
17:52 dbolser (after adding sequences)
17:53 dbolser (upon trying to retrieve a sequence)
17:58 rbuels dbolser: you probably have differing perl lib paths with different storables
17:58 rbuels dbolser: maybe you did these on different machines
17:58 dbolser I loaded features on box x, queried on box y, everything fine
17:58 dbolser I then loaded sequences on box y, queried on box y, problem emerged
17:58 dbolser I'll try loading everything on box x
18:04 carandraug joined #bioperl
18:05 dbolser sorry for the reformatting... here comes the pull request...
18:05 dbolser my commit message is bad
18:18 dbolser I'll let you know if it works tomorrow
18:18 dbolser night
18:23 dbolser hey! it works :D
18:25 philsf joined #bioperl
18:25 philsf left #bioperl
18:25 philsf joined #bioperl
18:38 CIA-119 bioperl-live: Dan Bolser master * r86b59a0 / scripts/Bio-SeqFeature-Sto​re/bp_seqfeature_load.PLS :
18:38 CIA-119 bioperl-live: rewrote usage script
18:38 CIA-119 bioperl-live: Comment by rbuels: actually, it's more of a refactor and correction of
18:38 CIA-119 bioperl-live: the bp_seqfeature_load.PLS usage message. - http://bit.ly/fruSr8
19:15 carandraug left #bioperl
19:23 svaksha left #bioperl
19:25 CIA-119 bioperl-live: Chris Fields master * ra8e2560 / t/SeqIO/swiss.t : fix test fail from bad file name - http://bit.ly/f94Lsi
19:25 CIA-119 bioperl-live: Chris Fields master * rca5289e / Bio/SeqIO/embl.pm : comment out overzealous error checking, which caused some EMBL files to not parse, fixes #3158 - http://bit.ly/eulaEH
20:04 philsf left #bioperl
20:07 svaksha joined #bioperl
20:10 sl33v3 left #bioperl
20:12 sl33v3 joined #bioperl
20:25 CIA-119 bioperl-live: Chris Fields master * r13d4fc6 / scripts/Bio-SeqFeature-Sto​re/bp_seqfeature_load.PLS : make sure short params don't conflict, make them explicit - http://bit.ly/fiW2Yn
20:51 CIA-119 bioperl-live: Chris Fields master * rd3ae617 / Bio/DB/SeqFeature/Store/DBI/mysql.pm :
20:51 CIA-119 bioperl-live: enable fast loading with namespaces; empty tables were being generated
20:51 CIA-119 bioperl-live: that had concatenated table names via _qualify() (e.g.
20:51 CIA-119 bioperl-live: volvox_volvox_feature); this fix checks for the presence of the
20:51 CIA-119 bioperl-live: namespace just in case and returns it w/o modification if it is found - http://bit.ly/hmFexg
20:54 sheenams joined #bioperl
22:02 sheenams left #bioperl

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