Camelia, the Perl 6 bug

IRC log for #bioperl, 2009-10-28

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

All times shown according to UTC.

Time Nick Message
01:48 ptl If I can get help here by being patient, does it mean that to give help I must be a doctor?
01:49 * rbuels opens ptl's cranium with a bone saw
01:49 rbuels nurse!  i need 20ccs of phenobarbitol stat!
01:49 rbuels not for him, dammit, for me!
01:51 ptl :P
01:58 ende joined #bioperl
05:50 ende joined #bioperl
12:09 rbuels joined #bioperl
13:53 kyanardag_ joined #bioperl
14:13 deafferret UUGHHH!! ME PATIENT!
15:08 * rbuels pulls deafferret's strings to make him look
15:08 rbuels crap!
15:08 rbuels lool!
15:09 rbuels lool!
15:09 rbuels stupid forearm muscles think they know what i want to type
15:09 * rbuels stabs his forearm with a pencil
15:10 deafferret assuming you use your a qwerty keyboard I hope you stabbed your right forearm
15:11 deafferret my stab-with-left-arm skills are not up to standards
15:11 deafferret and my left foreharm hurts from hours of swamp biking yesterday
15:11 deafferret typing 'qwerty' is especially painful
15:38 brunov joined #bioperl
16:08 DanC I'm writing code to grab Pfam/SCOP/CATH residue coordinates from SIFTS data, but struggling with coordinate systems. Anybody here familiar with SIFTS? http://www.ebi.ac.uk/pdbe/docs/sifts/
16:10 rbuels DanC: probably should try asking on the mailing list if you haven't already
16:10 rbuels DanC: there are more people there
16:10 DanC yeah, but I'd have to join the mailing list, and I have an aversion to that. I guess I should get over it.
16:11 rbuels DanC: you don't have to join it to send an email to it
16:11 rbuels DanC: people will send responses directly to you, usually
16:11 rbuels DanC: as well as CCing the list
16:11 DanC oh? I tried sending to a biopython list and it bounced with "only subscribers can post". I assumed the bioperl lists were configured similarly.
16:13 brunov so what's the latest word these days? Is there a plan to split things up a little more in BioPerl?
16:23 rbuels brunov: yes, there is, it's just a matter of doing it
16:23 rbuels brunov: so far there has not been much traction on it
16:23 brunov rbuels: any way I can helpl?
16:23 brunov *help
16:23 rbuels absolutely.
16:23 * rbuels looks on the gff page
16:24 brunov because I would *really* like for it to get done at some point, but don't know where or how to start
16:24 rbuels have a look at this page: http://www.bioperl.org/wiki/GFF_Refactor
16:24 rbuels this is the part that i'm working on
16:25 rbuels so, what needs to be done for this corner of it is to split FeatureIO off into its own distribution
16:25 rbuels which i've already mostly done, in an svn branch
16:25 rbuels you know how to use branches in svn?
16:25 brunov no, but I've heard of git-svn
16:25 rbuels lol
16:25 brunov I may have to take a look at it
16:26 rbuels git-svn isn't going to cut it for this
16:26 rbuels unfortunately
16:26 brunov oh
16:26 brunov well, then I'll have to learn the svn way
16:26 rbuels since the project is still mired in svn
16:26 rbuels you might be able to use svk to do it
16:26 rbuels svk is kind of a predecessor to git
16:26 rbuels and it uses svn as a backend
16:27 rbuels so that would be compatible with the current svn repos
16:28 brunov ok. So first I'll read up on the GFF problem, it's a good start thanks
16:29 brunov because somehow it feels wrong to have to depend on the whole of bioperl just to get a fasta parser
16:29 rbuels well, actually probably the best thing for you to start on would be finding one of the low-hanging pieces that can be split off
16:29 * rbuels looks for the email....
16:30 brunov Well, the trivial thing would be to split Bio::Root:: first, and then continue from there, right?
16:30 brunov just thinking out loud though
16:30 rbuels hmmmm
16:31 rbuels yes actually i suppose you could probably split from the bottom up as well as the top down
16:31 rbuels i don't think anybody has suggested that before actually
16:31 rbuels people have mostly been talking about splitting from the top down
16:31 rbuels splitting of featureIO is top-down for example
16:32 rbuels but it could also go in the other direction
16:32 brunov Well but since Bio::Root doesn't depend on anything else than itself, then it should be feasible...
16:32 rbuels to split bio::root, you would make a new distribution, and add a dependency on it in bioperl-live and the other monolithic distros
16:33 brunov right
16:33 rbuels this might be a good thing to suggest on the mailing list
16:33 brunov and then maybe see if you can untangle cross-dependencies among second-level namespaces and split them there
16:33 brunov like, having Bio::Seq as one dist that just depends on Bio::Root, and so on
16:34 rbuels there's also splitting from the top down, which has been the main strategy that people have been talking about
16:34 brunov yes, I read about how they split the GD modules, that was great
16:34 brunov because those are always a pain to install
16:35 rbuels brunov: read this thread: http://bioperl.org/pipermail/b​ioperl-l/2009-July/030573.html
16:36 rbuels also: http://bioperl.org/pipermail/b​ioperl-l/2009-July/030601.html
16:36 brunov good read, I just started reading your response to the maj's first mail, and I agree wholeheartedly
16:38 rbuels unfortunately, since then, not much traction on it
16:38 rbuels i started doing some work on featureio, but i got a big heap of other stuff land on my plate that pushed it asidef
16:39 brunov yes, I understand
16:39 rbuels you can use git-svn once you actually make a branch with svn, actually
16:39 brunov ok, so I'll catch up with all this reading, I'll try to straighten my thoughts on it, and see if I can find a way to help in this thing
16:39 rbuels that's kind of what i do
16:40 brunov wrt the mailing list, feel free to comment on it yourself if you want, I'm not yet confident enough about what I think to tell everyone
16:41 rbuels brunov: see Jason Stajich's mail in the bioperl reorganization thread for some ideas on things that might be easy to split off
16:41 rbuels brunov: if i were you i would pick one of these and try to do it
16:41 brunov rbuels: sounds great
16:41 rbuels there are some other ideas in that thread as well
16:41 rbuels for splitting off
16:42 brunov ultimately, it would be nice to say 'cpan Bio::Seq' and have just a Bio-Root and Bio-Seq dist get pulled in
16:43 rbuels yep, that's the idea
16:44 rbuels it's just a matter of untangling all the dependencies
16:53 deafferret oh. so Bio::Seq gone out of trunk is intentional?
16:53 rbuels lol
16:53 deafferret I hit that yesterday, thought someone typod something, didn't have time to chase  :)
16:53 * deafferret deafferret stopped using bioperl-live for the first time ever yesterday  :D
16:53 * deafferret deafferret deafferret deafferret !
16:53 brunov deaferret!
16:54 brunov *deafferret!
16:54 deafferret mmmm Brady Bunch references
16:54 * deafferret wants to be an array
16:55 deafferret mmmm arrays
16:55 * deafferret is a plurality once again
16:56 brunov I like hashes
16:56 brunov and turtles
16:59 deafferret rbuels is hiding Bio::Seq secrets in his brain
16:59 * deafferret starts drilling
17:05 * ptl starts drolling
17:05 * ptl starts droolling
17:06 deafferret maybe ptl is hiding the same secrets?
17:07 ptl no, it's my zombie side hungry for your brains...!!
17:08 deafferret so if I asked you why Bio::Seq isn't in bioperl-live trunk you'd stare at me blankly?
17:08 deafferret I'm too lazy to do any real research, you see
17:08 deafferret IRC... mmm... lazyweb
17:09 ptl My zombie side can't talk knowingly about live stuff :(
17:09 ptl if there was a bioperl-dead trunk maybe it could.
17:10 rbuels deafferret: wtf are you talking about.  Bio::Seq is there
17:10 deafferret oh ya? don't make me check again
17:10 deafferret FUD!! F U D ! !
17:10 rbuels i'm looking at my bioperl-live svn head right here
17:12 deafferret grrr   svn: Decompression of svndiff data failed
17:12 rbuels lol
17:12 rbuels what is wrong with your svn you weirdo
17:16 rbuels deafferret: http://subversion.open.collab.net/ds/viewMe​ssage.do?dsForumId=4&dsMessageId=335760​&orderBy=createDate&orderType=desc
17:16 rbuels oddly-configured firewall possible?
17:17 rbuels deafferret: does this happen when you are on a different network?
17:20 deafferret rbuels: yup. just hit the same error from entirely different network
17:20 deafferret & lunch
17:43 pyrimidine joined #bioperl
17:44 pyrimidine Anyone here testing out bioperl on hadoop (amazon ec2, etc)?  I know mark jensen's got something up on amazon.
17:45 pyrimidine I'm likely doing an interview for FLOSS Weekly with Jason on BioPerl and this is something that's popped up.
17:45 pyrimidine (hadoop or 'cloud', either way)
17:45 rbuels pyrimidine: that's interesting.  how does randall do his interviews, does he come to you?
17:46 pyrimidine asked out of the blue.  jason was approached first
17:46 pyrimidine he asked me to join him
17:46 rbuels pyrimidine: but i mean is he going to physically sit there with you guys or what
17:46 pyrimidine nope, via skype
17:46 rbuels ah, ok
17:47 pyrimidine I'll be using hadoop here (I have access to the yahoo cloud), but I'm checking around to see if anyone else has tested it out
17:48 rbuels pyrimidine: haven't used either of those but i'd love to hear about what you do with them
17:49 pyrimidine one example: http://www.slideshare.net/tracylaxdal/aws-start​up-event-washington-dc-2009-schatz-cloud-burst
17:50 pyrimidine paper: http://www.ncbi.nlm.nih.gov/pubmed/19357099
17:50 pyrimidine we may have something local that's faster, but I need to test it out
17:51 pyrimidine (possibly run a next-gen rnaseq mapping for around $10 on Amazon EC2)
17:51 rbuels so how do you spend grant money on that?
17:52 pyrimidine very easily ;)
17:52 rbuels i doubt that.  wouldn't it need to be in the grant budget or somethign?
17:53 pyrimidine yes. allocate to an account and use a dedicated purchase order card
17:53 rbuels sigh
17:53 rbuels figured as much
17:53 rbuels but if you want to buy hardware that has to be done similarly too
17:53 rbuels so i guess it's not really much more hassle
17:54 pyrimidine right.  comes down to cost; $10-100/run on the cloud, or $50K-100K for a small cluster (and a longer wait)
17:55 rbuels seems that the cloud would win 90% of the time
17:55 pyrimidine downside: larger learning curve.  Hadoop is not like SGE Rocks or similar, and is primarily for batch.
17:56 rbuels it's java-specific too isn't it?
17:56 pyrimidine but one could automate some analyses (upload, run, cleanup, download results)
17:56 pyrimidine not if you use hadoop-streaming
17:56 * rbuels looks up this hadoop-streaming you speak of
17:56 pyrimidine http://wiki.apache.org/hadoop/HadoopStreaming
17:57 pyrimidine can use perl, python, shell, etc.  Need to stipulate a mapper and reducer (or just a mapper)
17:58 rbuels so it uploads the mapper and reducer binaries to the cloud?
17:58 rbuels and runs them?
17:58 pyrimidine yep
17:59 pyrimidine python example: http://www.michael-noll.com/blog/2007/09/21/tutori​al-writing-an-hadoop-mapreduce-program-in-python/
18:00 pyrimidine we had a month-long indexing job here that ran in ~10 hrs on our local hadoop
18:00 pyrimidine probably could cut it down to 2-5 hrs
18:06 rbuels and yahoo sells access to a hadoop cluster as a service?
18:08 pyrimidine no.  I have access to a hadoop testbed (donated by yahoo).  uiuc only, afaik
18:09 rbuels ah.
18:09 pyrimidine though there are others who offer cheaper clouds than amazon.  I think IBM has a cheaper option.
18:20 ende joined #bioperl
18:46 brunov \o/
18:47 deafferret so no response on the mailing list. So I must be the only one seeing this. From two different networks, 2 different OSs...  svn: Decompression of svndiff data failed
18:48 rbuels deafferret: even on a clean checkout?
18:48 deafferret rbuels: I'm going to send that ICMP link you sent to our admin guy and see if that makes any sense to him
18:48 deafferret rbuels: I'm doing a clean checkout and hit that. yes.
18:48 deafferret two different networks.
18:48 deafferret two different OSs. I'm amazed it's suddenly happening to just me
18:49 deafferret svn co svn://code.open-bio.org/bioperl/bioperl-live/trunk bioperl-live-new
18:49 deafferret not ssh
18:49 rbuels from every single network?
18:49 rbuels home?  work? coffee shop?
18:49 deafferret @work[0,1]
18:49 deafferret have not tried the other 7trillion networks yet
18:50 rbuels but not on other machines right?
18:50 rbuels and is this recent with the snow leopard upgrade?
18:50 deafferret Linux server on one network. My laptop on two different networks so far
18:50 deafferret yes, 10.6.1 up to date laptop
18:50 rbuels so you're getting it from a linux server too???
18:50 deafferret yes
18:51 deafferret let me try another linux server
18:51 pyrimidine try it from dev.open-bio.org
18:51 deafferret pyrimidine: o-tay
18:52 deafferret 2nd linux server just failed
18:52 rbuels oops, i just hit it too
18:52 deafferret FFFF  YA!
18:52 deafferret HA!
18:52 rbuels ok it's probably something on the server end
18:53 rbuels svn: Decompression of svndiff data failed
18:53 pyrimidine here too
18:53 * deafferret victory dances (clothes on)
18:53 rbuels svnadmin check?
18:53 ende I got that error too
18:53 pyrimidine want me to report it to support?
18:53 deafferret yes please
18:55 deafferret pyrimidine: they might check for them blocking a specific type of ICMP
18:55 deafferret http://subversion.open.collab.net/ds/viewMe​ssage.do?dsForumId=4&dsMessageId=335761
18:55 pyrimidine deafferret: ok, I'll add that to the report
18:55 deafferret pyrimidine++
18:55 deafferret "svnadmin verify" apparently
18:55 deafferret (I've never run that.)
18:56 ende pyrimidine, you have a bot in this chan for github ?
18:56 pyrimidine just the standard github bot, for biome and bioperl6
18:57 pyrimidine I plan on setting up something more dedicated when I set up my server (next week or two)
18:57 ende confused, does that report in here when commits are made to the main bioperl svn repos?
18:57 pyrimidine nope
18:57 pyrimidine just github for now
18:58 pyrimidine we could add a post-commit hook for bioperl svn commits
18:58 deafferret pyrimidine: is there an up-to-date github I can clone now? (Since svn is broken? -grin-)
18:59 rbuels hrrmm
18:59 pyrimidine you have a dev account?
18:59 deafferret I have an dev.open-bio.org account. Is that it?
18:59 pyrimidine yep.
18:59 pyrimidine http://www.bioperl.org/wiki/Using_Su​bversion#Checking_out_code_from_the_​repository_with_a_developer_account
19:01 deafferret pyrimidine: hmm? you just mentioned 'just github for now' which I thought meant there was a github of bioperl-live now
19:01 pyrimidine no, just biome and bioperl6.
19:01 deafferret ah. roger. thanks.
19:03 pyrimidine so, do we want to pollute-up the channel with bioperl-related svn commits?  Or leave them off?  We do have bioperl-guts for those so inclined...
19:04 * deafferret votes for a 1-line message in here on every commit
19:04 * deafferret votes for deafferret not implimenting that
19:04 pyrimidine I can inquire about #perl6's dalek for those purposes.
19:10 * rbuels votes for one-line message here on every commit, and for deafferret implementing that
19:10 * rbuels pulls deafferret's strings and raises his hand
19:11 rbuels *thunk thunk thunk* stop hitting yourself!
19:13 rbuels pyrimidine: did you read the logs and see what brunov said about Bio::Root?
19:13 rbuels he made an interesting point that splitting can go from the bottom up as well
19:14 rbuels you just have to add deps to things above it
19:15 deafferret wow. 17 nicks in here.
19:16 rbuels 4 nicks in #gmod!
19:16 rbuels truly these are heady days
19:20 pyrimidine pon
19:20 pyrimidine *pong
19:21 pyrimidine ENOBRUNOV
19:21 brunov joined #bioperl
19:24 pyrimidine rbuels: I tend to agree re: splitting things up.  I think a bottom-up approach may be easier in some respects.
19:24 pyrimidine rbuels: best proof is to get something working, maybe on a branch, then release it on the unsuspecting public
19:26 rbuels i hadn't thought of the possibility of splitting from the bottom as well as the top until brunov mentioned it
19:26 brunov at leat for Bio::Root, it seems doable. I took a look, and looks pretty much self-contained
19:26 brunov things get a liitle bit hairier from there up though
19:27 rbuels yeah it's kind of like chipping away at a big rock
19:27 brunov but maybe if it's done with Bio::Root, the next low hanging fruits would be modules that *only* depend on Bio::Root
19:27 brunov or peeling an onion
19:27 brunov :)
19:32 pyrimidine brunov: sounds fine by me.  One could always take this to a branch and split out the relevant bits.  The tricky bit is how one goes about initially testing this prior to a CPAN launch.
19:33 pyrimidine other is convincing the gmod folks to update their net-install to handle this
19:33 pyrimidine (remember, it has a -d option for installing via svn)
19:34 rbuels wait, gmod netinstall installs bioperl????
19:34 * rbuels slaps his forehead
19:34 pyrimidine yes
19:35 rbuels you mean the gbrowse net installer?
19:35 pyrimidine yes
19:36 pyrimidine default is install via CPAN, but it can download bioperl-live via svn and install using the 'developer' option
19:37 rbuels ah, that's a reaction to the slow release cycle of the past(?)
19:37 pyrimidine prior to a deconstructed bioperl hitting cpan, we would need to get things like that working correctly
19:37 pyrimidine yes
19:37 rbuels i'm one to talk, not having done jack
19:38 pyrimidine ah, but development is ramping up a bit.
19:38 rbuels a bit.
19:38 pyrimidine 1.6.1 - 1.6.0 was 9 months
19:38 pyrimidine 1.6.0 - 1.5.2 was around 2 yrs
19:38 rbuels heh.
19:39 pyrimidine 1.5.2 - 1.5.1 roughly the same
19:39 rbuels how do you have so much time to devote to bp?
19:39 pyrimidine (2 yrs)
19:39 * rbuels has been struggling with sgn internal code lately, completely reorganizing it
19:39 pyrimidine lately, I don't have as much time.
19:40 pyrimidine but, I make it a priority to try incorporating it into daily work, which pushes development forward
19:41 pyrimidine pretty soon you'll see me popping up on gmod/gbrowse, as well as ensembl
19:41 rbuels and your daily work has a very large overlap with the core bp features
19:41 * rbuels nods
19:41 pyrimidine yes.
19:41 rbuels makes sense
19:41 * rbuels points at #gmod
19:42 rbuels oh
19:42 rbuels you figured that out 15 minutes ago
19:42 rbuels nm
19:42 rbuels lol
19:42 pyrimidine np
19:43 pyrimidine $job also involves trying to make lightweight objects, so I'll be pushing biome a bit more to lazily load sequence data, prob. via a db handle
19:44 pyrimidine this of course will probably trickle back into bioperl if it's feasible
19:45 rbuels and breaking up the bioperl monolith would greatly facilitate that, since it would make it much easier to justify additional dependencies (like on moose)
19:45 pyrimidine yep
19:45 * rbuels is going over familiar territory now
19:45 brunov agreed
19:45 pyrimidine we already have someone (of-list) working on a hmmer3 parser that uses Moose
19:45 pyrimidine *off-list
19:46 pyrimidine you'll probably see that popping up on bioperl-dev soon, just to keep the dependencies down in core
19:47 brunov why not start off my making it a separate dist?
19:48 * rbuels paints the bike shed blue
19:48 pyrimidine That's a possibility as well, which I had mentioned.
19:48 brunov oh, sorry
19:48 rbuels i almost said that too brunov
19:48 rbuels lol
19:49 pyrimidine The problem (specific to hmmer3) is the output may be in flux (it's still alpha)
19:49 rbuels (the separate dist thing)
19:49 rbuels anyway gents, i need to get back to work
19:49 pyrimidine same here
19:50 rbuels maybe once i finish this damn reorganization (which is about 2 months behind schedule), i can get back on the featureio/typedseqfeature stuff
19:50 rbuels &
20:41 pyrimidine commuting &
21:39 brunov joined #bioperl
21:41 brunov rbuels: ping
21:41 rbuels brunov: why?
21:42 brunov Have any of you had problems checking out the svn repo?
21:42 brunov i'm getting a diff error
21:42 rbuels yeah they're working on it
21:42 brunov oh, so it's not me
21:42 rbuels it's something on the open-bio server
21:42 brunov oh, ok
21:42 brunov darn
21:42 rbuels they need to track down one of the sysadmins
21:43 brunov wrt bioperl diet, Bio::Structure and Bio::Biblio should be also pretty straightforward to weed out, right?
21:43 rbuels i don't know, i've never used either of them
21:43 rbuels Module::Depends might help you evaluate that
21:43 rbuels or maybe ExtUtils::Depends
21:43 brunov oh, nice
21:44 brunov I tried to use Class::Sniff but it couldn't cope with it
21:44 rbuels nm about extutils::depends
21:44 rbuels ah, m::d won't work
21:44 brunov oh, right
21:44 rbuels i had something.....
21:44 * rbuels looks
21:44 brunov it reads meta.yaml
21:45 brunov oh, crhis's dependencies.pl, right?
21:45 rbuels there's that
21:45 rbuels i also had something though
21:46 rbuels hrm, my thing was just using regexps
21:47 rbuels run( command => [qw/grep -hP/, qq@^\\s*(use|require)\\s+.+;@, @{$k->{perlfiles}}],) is the money line
21:47 rbuels grep -hP '^\\s*(use|require)\\s+.+;'  *.pm
21:47 rbuels gets most of em
21:48 brunov O.o
21:48 * rbuels uses grep cause it's faster for special cases like this
21:49 * rbuels looks at pyrimidine's thing
21:49 rbuels oh, it's the same thing
21:50 rbuels except it skips Bio::
21:50 rbuels i think
21:50 * rbuels doesn't really remember so will shut up now
21:50 rbuels i have to go do some errands, back in an hour or two
21:50 rbuels &
21:51 brunov k, bye

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