Camelia, the Perl 6 bug

IRC log for #bioclipse, 2008-02-01

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

All times shown according to UTC.

Time Nick Message
07:53 EskilA joined #bioclipse
08:06 Annzi joined #bioclipse
08:22 olas joined #bioclipse
08:37 de_maas joined #bioclipse
08:55 edrin joined #bioclipse
08:55 edrin moin
08:59 egonw_ joined #bioclipse
09:13 olas moin all
09:13 olas http://www.medchem.leidenun​iv.nl/home/coffeetrial.htm
09:15 olas anyone going to: http://www.int-conf-chem-structures​.org/html/general_information.html ?
09:15 EskilA joined #bioclipse
09:22 egonw_ olas: might go
09:23 egonw_ not sure yet, want to, but depends on how travels funds are left...
09:23 olas sure
09:23 olas I don't know if I will go
09:23 olas but the topics are interesting
09:23 egonw_ it's for chemoinformatics the best conference there is
09:23 olas then you should go!
09:24 edrin regarding travel funds... i still did not request my travel flight funds for the bioclipse meeting last year :/
09:24 egonw_ well... from a financial point of view, I'm a *bio*informatician now ...
09:25 edrin do you think I can still request them?
09:25 olas edrin: I don't know
09:25 olas send an email to erik bongcam
09:26 olas you can at least ask
09:26 olas Erik.Bongcam@lcb.uu.se
09:27 edrin yes, thank you
09:29 olas egonw_: Could not Christoph fund you on behalf of cdk?
09:29 egonw_ no, don't think so
09:29 olas you do a lot of work on cdk...
09:29 olas ok
09:29 egonw_ yeah, but not for him
09:29 olas how are things progressing with MetWare?
09:29 egonw_ quite good
09:29 egonw_ the first modules are going online now
09:29 egonw_ MetID
09:29 egonw_ and MetRAW
09:30 olas Can we build a client for Bioclipse?
09:30 egonw_ that would be great
09:30 olas would it be hard?
09:30 egonw_ no, don't think so
09:30 egonw_ the idea could be like this:
09:30 jonalv joined #bioclipse
09:30 egonw_ you create a BC2 database
09:30 egonw_ and then import data from MetWare
09:30 egonw_ which would be using the BioMart interface
09:31 egonw_ which would also mean...
09:31 egonw_ direct compatibility with lot's of bioinformatics databases...
09:31 olas what type of data would you import?
09:31 egonw_ all it needs is a "Import from BioMart" functionality
09:31 olas and how would the BC2 database look like?
09:31 egonw_ BioMart had one type of object central (star design)
09:31 olas I've been thinking of Biomart plugin for a long time
09:31 egonw_ with properties around it
09:32 egonw_ these may be spectra (MetOBSERV), molecules (MetID)...
09:32 olas so how would teh design of a BC2 DB look like?
09:32 egonw_ or sequences (Ensemble at EBI)
09:32 egonw_ which uses BioMart too... (well, they developed it :)
09:32 olas I don't quite follow
09:33 edrin what is Christoph going for projects in EBI now?
09:33 egonw_ see biomart.org
09:33 olas egonw_: yea, but how would you envision a BC2 DB to look like?
09:33 olas we don't want a local biomart, right?
09:33 egonw_ depends...
09:33 egonw_ a list of sequences
09:33 egonw_ a list of molecules
09:34 olas list?
09:34 egonw_ right, no local BioMart
09:34 egonw_ [10:31] <egonw_> all it needs is a "Import from BioMart" functionality
09:34 olas but import what in what format?
09:34 olas sequences: right
09:34 olas molecules: right
09:34 egonw_ BioMart result format
09:34 edrin hm, egon, is it like a java lib for p2p "file" transfere?
09:34 egonw_ depends on what it in the database...
09:34 olas which is?
09:34 egonw_ sequences -> string
09:35 egonw_ molecules -> MDL molfile/CML
09:35 olas ok, so it's light-weight
09:35 jonalv so we would store a list of molecules and a list of sequences?
09:35 egonw_ edrin: no, it's like a query interface to access large databases
09:35 jonalv very light weight
09:35 olas good
09:35 egonw_ olas, jonalv: no
09:35 egonw_ olas, jonalv: biomart does not put any restrictions on hwat is in there
09:35 egonw_ see: metware.sf.net
09:36 egonw_ for the full complexity we are using
09:36 egonw_ however...
09:36 egonw_ the import might be light weight
09:36 olas yea
09:36 egonw_ biomart is the interface between data and analysis
09:36 olas we need to decide what we want to import
09:36 jonalv so our database system will link out to bigger systems when online?
09:36 egonw_ you don't want to full db locally mirrored...
09:36 egonw_ you want to select some of its data, and analyze that
09:37 egonw_ olas: only that related to the IBioResource's BC2 has
09:37 egonw_ so, list of molecules
09:37 olas but import molecules in MDL and sequences in some format (EMBL, FASTA) will be light-weight and quite easy
09:37 egonw_ list of sequences
09:37 egonw_ proteins
09:37 olas yes
09:37 egonw_ but damn useful
09:37 olas egonw_: we will investigate setting up a BioSQL database for managing sequences.
09:37 olas it's already there in BioJava
09:38 egonw_ sounds good
09:38 olas and for molecules: any suggestions?
09:38 olas Else we will build a light-weight db as discussed before
09:38 egonw_ couple of possibilities...
09:39 olas id, name, SMILES, INCHI, and content as CML
09:39 olas fingerprint as well I guess
09:39 jonalv what operations would our light weight database support?
09:39 egonw_ jerome's MySQL db, sMol, the one used by the people doing the screening for REACH (ambit), or the model from nmrshiftdb.org
09:40 olas egonw_, jonalv: We should have a wiki page with use cases of what we want to be able to do
09:40 olas jonalv: could you not start to set that up and egonw_ might add things?
09:40 olas and me
09:40 egonw_ most important use case:
09:40 jonalv is <http://wiki.bioclipse.net/i​ndex.php?title=Structure_DB> already dead then?
09:40 egonw_ "Import from MDL SD file"
09:40 egonw_ and simply browse the content
09:41 egonw_ de_maas: do you want to add anything to that?
09:41 olas jonalv: no
09:41 olas I still think it's the best
09:41 olas the easiest for users
09:41 egonw_ oh, btw... did you see the 'featured articles' this week of BMC Bioinformatics...
09:41 jonalv good because I have the DAO layer working in that structre...
09:41 egonw_ check out the screenshots :)
09:41 olas doi?
09:41 olas oh
09:41 olas sorry
09:42 egonw_ no, worries... time for a sauna
09:42 jonalv sry?
09:43 olas http://www.biomedcentral.com/bmcbioinformatics/
09:43 olas I like it a lot :-)
09:43 egonw_ good :)
09:43 egonw_ same here, btw
09:43 olas egonw_: looking forward to having scripts on web pages run inside of Bioclipse
09:44 olas that will be powerful
09:44 egonw_ quite so
09:44 jonalv what is it I should be looking at?
09:44 egonw_ you saw that recent paper, where they do something like that, right?
09:44 egonw_ based on Gaggle, I think
09:44 olas yes
09:44 olas I have had the same idea
09:44 egonw_ yeah, it's in the air
09:45 olas but will be so much mor epowerful inside Bioclipse
09:45 egonw_ jonalv: check the screenshots
09:45 olas you can add new parsers/scripts via an EP
09:45 egonw_ jonalv: of the 'Featured Articles' section
09:45 egonw_ ah, cool...
09:45 jonalv ah
09:45 egonw_ a net.bioclipse.userscripts plugin :)
09:45 egonw_ olas: good thinking
09:46 olas along those lines
09:46 olas you can contrivute the script to extract data, and views to visualize them
09:46 olas for example
09:46 egonw_ though you would probably also need a userscript.manager plugin, where people can register/edit custom scripts too
09:46 olas yup
09:46 olas but since we have full JS support in Bioclipse
09:46 olas you can write scripts on the fly
09:47 jonalv would this be working with the built in web browser in Eclipse?
09:47 olas we would probably use mozilla engine in Bioclipse
09:47 jonalv Isn't that running naively? Can we acces it on all lpatforms?
09:47 olas Mozilla engine is available for Eclipse
09:48 olas The builtäin browser is not so nice...
09:48 jonalv but we can run mozilla within Bioclipse?
09:49 jonalv cool
09:49 olas I have read it
09:49 olas but not tried it
09:49 egonw_ oi
09:49 egonw_ spam in blogs is bad
09:49 jonalv okey but it looked promising?
09:50 egonw_ but bloody anyone when you are registerd to receiving comments on a comment you made in another blog, and that item gets spammed :(
09:50 jonalv hm sad
09:50 egonw_ bloody anyone -> bloody annoying
09:50 jonalv ah
09:51 olas jonalv, egonw_: how about putting use cases here? http://wiki.bioclipse.net/index.php?tit​le=Bioclipse_databases&amp;action=edit
09:51 egonw_ olas: see http://wiki.bioclipse.net/index​.php?title=Bioclipse_databases
09:53 olas :-)
09:54 egonw_ jonalv: your page is very nice
09:54 egonw_ gives a good idea of what it should look like
09:54 egonw_ but doesn't discuss use cases...
09:55 egonw_ shall I quickly hack sometihng up for that new page?
09:55 jonalv egonw_, no it doesn't... I just explained what I had done so far...
09:55 jonalv please do, I'll check it out
09:56 jonalv I didn't quite follow though. Will we use the structure I started with structuredb?
09:58 jonalv The database and dataaccess objects are more or less working so it might be something worth buildin upon. Unless we want to set up some big database system that's already designed and working...
10:02 olas I think we should have simplicity
10:03 olas that's why we started your database project
10:03 egonw_ almost done typing up that page
10:03 olas it should be EASY for users to create and manage a small database
10:03 olas relying on external DB's is hard
10:04 olas we can always add support for mor ecomplex databases as well, but that will be an additional feature
10:04 jonalv yes and if users want a big database that is probably somthing that we shouldn't start with...
10:04 jonalv good
10:05 jonalv but use cases are important otherwise we might end up with a too light weight system not being able to do much at all...
10:05 egonw_ http://wiki.bioclipse.net/index​.php?title=Bioclipse_databases
10:05 egonw_ there's a start
10:08 jonalv egonw_, wow you write fast :)
10:13 edrin bc2 is supposed to be run with Java SE 6, right?
10:14 jonalv edrin, not what I know. Why?
10:14 jonalv IS there even a Java 6 for the mac users?
10:14 edrin dunno
10:15 jonalv I know masak had problems with that... But I don't know how he solved them... olas?
10:15 edrin just wondering if we may use libs coming with > Java SE6
10:15 jonalv what lib would you like to use? something about web services?
10:16 edrin no
10:16 edrin scripts
10:16 edrin java scripts
10:16 jonalv ah okey
10:16 olas preferably not
10:16 olas but maybe
10:17 jonalv what does it do? (curious)
10:17 edrin nothing yet
10:17 olas must be able to run evrywhere
10:17 jonalv the libs you want to use..
10:17 edrin just wondering if there is a possibility to depend on java 6
10:19 olas well, maybe
10:19 olas We can bundle the JRE with Bioclipse2
10:19 olas so in principle we can use any JRE we like
10:19 olas if it is stable on all platforms
10:20 jonalv But what are the benefits? What is it that we can't do now that we would be able to do?
10:20 olas (including Mac OS X)
10:25 Annzi left #bioclipse
10:28 edrin java SE 6 comes with scripting support API, it already includes javascript Rhino, in addition to that we could make use of https://scripting.dev.java.net/ - Rubym, Python, php... just have to be JSR 223 compatible. And it seams many are...
10:29 edrin however i would recommend not to use too many scripting engines in parallel. because you can't learn a new scripting lang each new day...
10:30 Annzi joined #bioclipse
10:31 jonalv wel that can be used as an opposite argument as well...
10:32 edrin jup
10:32 edrin but look here!
10:32 edrin http://www.judoscript.com/books/judosc​ript-0.9/chapters/wsdl.html#wsclients
10:32 edrin isn't this cool?
10:32 edrin olas, egonw_: have you seen this: http://www.judoscript.com/books/judosc​ript-0.9/chapters/wsdl.html#wsclients ?
10:35 Annz1 joined #bioclipse
10:38 olas no
10:38 olas not seen it
10:39 olas what is Judo?
10:39 Annz1 left #bioclipse
10:39 olas masak should really take a look at that
10:39 jonalv mail it to him
10:45 Annz1 joined #bioclipse
12:35 CIA-21 bioclipse: ospjuth * r4874 /branches/bioclipse2/core/p​lugins/net.bioclipse.core/ (10 files in 3 dirs): Added convenience methods for Jobs in Bioclipse2.
12:37 CIA-21 bioclipse: ospjuth * r4875 /branches/bioclipse2/bioinformatics​/plugins/net.bioclipse.biojava.ui/ (4 files in 2 dirs): Added 2 long running operations to demonstrate failing and succeeding jobs. Right-click on an IAASequence to execute (e.g. expand sequence2.fasta in sampleData/sequences).
13:24 CIA-21 bioclipse: ospjuth * r4876 /branches/bioclipse2/bioinformatics/plug​ins/net.bioclipse.biojava.ui/plugin.xml:
13:24 CIA-21 bioclipse: Solved how actions can be visible/enabled for certain contentTypes. Example
13:24 CIA-21 bioclipse: implementation: Long running operations only available when clicked a File which
13:24 CIA-21 bioclipse: is considered a sequence (with SequenceContentType that matches the
13:24 CIA-21 bioclipse: file-extensions).
13:37 olas egonw_: What can I do if I get a "Checksum mismatch" in SVN? I Can't update nor commit...
13:39 egonw_ no idea
13:39 olas :-(
13:39 egonw_ did you do a good svn update
13:39 egonw_ ?
13:39 olas yes
13:39 egonw_ try this:
13:39 olas my commit broke in the middle
13:39 egonw_ make a fresh checkout...
13:39 olas had to close subclipse
13:39 egonw_ copy in your new files, and try again
13:39 egonw_ but first try to commit from the command line
13:39 olas do I need to do a fresh checkout?
13:40 olas damn, that's heave
13:40 olas heavy
13:40 olas sure, I could delete the plugin and check out from SVN
13:43 olas guess I have to do that :-(
13:43 olas damn, I hate subclipse
13:44 olas and the combo subclipse+sf
13:55 egonw_ olas: want to try a BioMoby service at PRI?
13:55 egonw_ MarkFiers managed to get one with Perl done
13:55 egonw_ where I/we failed with jMoby so far
13:59 MarkFiers joined #bioclipse
13:59 egonw_ olas: meet MarkFiers
13:59 MarkFiers Hi olas,
13:59 egonw_ MarkFiers: meet Ola Spjuth
13:59 MarkFiers Hi Ola,
13:59 olas hi MarkFiers
13:59 egonw_ olas: MarkFiers is the workflow expert here at PRI
13:59 olas nice to meet you
13:59 MarkFiers Nice to meet you too.
13:59 egonw_ a paper on his workflow work is about to appear in BMC Bioinfo
14:00 MarkFiers correct...
14:00 egonw_ olas: he is currently setting up BioMoby services
14:00 olas I see
14:00 MarkFiers I know your name Ola, but cant place you right away..
14:00 egonw_ MarkFiers: olas is the big buy behind bioclipse.net
14:00 olas have you seen the prototype page: http://wiki.bioclipse.net/index.php?​title=Run_Workflows_inside_Bioclipse ?
14:00 olas It's not operational yet though
14:00 MarkFiers not yet...
14:01 olas I have not had time to finish it
14:01 MarkFiers It;s taverna in bioclipse..
14:01 MarkFiers or am i wrong?
14:01 olas yes
14:01 egonw_ correct
14:01 olas that is the goal
14:01 olas to be ablt to run WF's from bioclipse
14:01 egonw_ MarkFiers has written a Taverna1 competitor
14:01 olas and mix and match with otehr functionality
14:01 MarkFiers I have not had the chance to look at bioclipse extensively
14:01 egonw_ that is... his system solves some T1 problems
14:01 egonw_ which T2 *might* or *might not* solve
14:01 olas you have a link to it?
14:02 MarkFiers http://www.ab.wur.nl/Cyrille2
14:02 MarkFiers not too much data there yet.. I can let yoy know when  the preprint is online at bmc
14:02 olas please do
14:03 olas I have worked a lot to make Bioclipse a client for BioMoby services
14:03 MarkFiers Ok, I will..
14:03 olas in Bioclipse1 there's a basic implementation
14:03 MarkFiers ok.. I will check it out.. that works fine now?
14:03 olas http://wiki.bioclipse.net/in​dex.php?title=BioMoby_plugin
14:03 olas but I'll make a good client for Bioclipse2 as soon as I get the time
14:04 MarkFiers Time seems to be a universal problem.
14:04 olas Mark: Did you write your own WF engine?
14:04 egonw_ which is why opensource and openscience project work so well :)
14:05 MarkFiers Yes, wrote my own. it is somehting which has grown slowly over the last few years, in the absence of systems which could do the sort of high throughput work we needed.
14:06 olas Is it java?
14:06 MarkFiers I must admit that I more or less had to kick out webservices to get it working. It was based on biomoby, and still is, but the services are gone.
14:06 MarkFiers No, python.
14:06 MarkFiers might get it working in jython, though, never tried.
14:07 olas ok
14:07 MarkFiers I like cyrill2 a lot, it is relatively stable, fault tolerant and implements provenance... so we can redo parts of a pipeline if necessary.
14:08 olas yes, that's important
14:08 MarkFiers and something i quite miss in taverna.
14:08 olas What do you think of Knime?
14:09 MarkFiers have not worked with it.
14:09 MarkFiers althoug egon gave met a cd...
14:09 MarkFiers haven't had enough time
14:10 olas :-)
14:13 MarkFiers Well, Ola, nice meeting you! I'll check out Bioclipse, or ask Egon for a demo.
14:13 edrin MarkFiers: a thing i always wondered and keep wondering is, is a completely graphical drag&drop workflow thing (like taverna) really something ppl need and will use?
14:14 egonw_ MarkFiers: oh...
14:15 egonw_ MarkFiers: edrin is the one behind those XMPP webservices
14:15 edrin for workflow users wont it be more suiteable to have a more "basic" implementation? like a friendly scripting IDE? with database that is simeple to use to keep&handle data?
14:15 egonw_ edrin: MarkFiers is the one why was interested in the technology
14:15 MarkFiers ok...
14:15 edrin ah, that's what I expected ;)
14:15 CIA-21 joined #bioclipse
14:15 MarkFiers now i remember... yes i like the technology!
14:16 MarkFiers i think drag&drop workflows are great in their own realm.
14:16 edrin I think prototyp can be done now
14:16 MarkFiers for (relatively small??) experimentt, prototyping and the like.
14:17 MarkFiers and depending on the user, a perl guru might prefer scripting, but the average bench biologist probably won't
14:17 MarkFiers don't you agree? Or do you have another opinion?
14:18 edrin yes, I agree
14:18 edrin A bench  Biologist wont use a programing language
14:18 edrin however
14:18 MarkFiers no, i don't think so..
14:19 MarkFiers brb
14:19 edrin hu?
14:19 egonw_ brb = be right back
14:20 edrin yes, but he does not think what he just thought?
14:20 egonw_ ummm... I think he agrees...
14:20 egonw_ typical to Dutch language...
14:20 MarkFiers ??
14:20 egonw_ when you confirm a negative statement, you say 'no'
14:20 MarkFiers i meant.. be right back..
14:21 egonw_ [15:18] <MarkFiers> no, i don't think so..
14:21 edrin lol
14:21 egonw_ you meant "yes, I don't think they will either..." .. right?
14:21 MarkFiers i was agreeing with you... must be dunglish
14:21 edrin heh
14:22 edrin Mark, what I just thought, the people would not use a Taverna, it is much too complicate to use, much too buggy...
14:22 edrin would not use... either
14:23 edrin the webbrowser thing is probably better
14:23 MarkFiers I agree that, at the moment, it is rather awkward to use, but i like the general idea
14:23 edrin yes
14:24 edrin i would prefere to just have a Wizard to generate a workflow.
14:24 MarkFiers a webbrowser can have the same problems, a badly designed interface is a badly desinged interface.
14:24 MarkFiers do you know myexperiment.org???
14:24 MarkFiers I like the idea a lot.
14:24 MarkFiers you can store complete workflows there
14:24 MarkFiers or snippets.
14:25 MarkFiers And, obviously, I have to mention my own system, Cyrille2, Y.A.W.E (Yet another workflow engine), also web based, and very much geared towards very high throughput workflows.
14:26 edrin Like "what do you want to do?" A: get certain amount of data from the web B: load data from files ... NEXT: "what do you want to do with the data?" A: run a test on them (Web Service) B: convert them ... NEXT: ...
14:26 edrin MarkFiers: yes, i looked at your webpage
14:26 MarkFiers good.
14:26 MarkFiers paper is due to be published any day now
14:27 edrin fine :) congrats
14:27 MarkFiers thnx :)
14:27 egonw_ MarkFiers: just for fun... tell them how much data typically goes through the engine right now
14:27 egonw_ the sequence data...
14:27 egonw_ to give them an idea
14:28 MarkFiers we do mainly genome annotation now, it typically handles hundreds of bacs, without complaining..
14:29 MarkFiers and we run anywhere from 10 to 60 analyses, resulting in millions of objects..
14:29 edrin cool
14:29 MarkFiers well, I like it :)
14:29 MarkFiers obviously
14:29 edrin what kind of tests do you run?
14:30 MarkFiers mainly gene precition, de-novo or similarity based (sim4, genewise). Blast, interpro,
14:30 edrin i mean on the gens in example
14:30 edrin ah ok
14:30 edrin and do you do also subcellular localisation predictions?
14:31 MarkFiers Isn't that partly in interpro?
14:31 edrin i don't knwo
14:31 edrin know
14:32 MarkFiers I thought signalp was in there, but it doesn't look that way.. so, i don't thing we doing subcell. localization... that would be a good addition
14:33 MarkFiers edrin: you are the one behind xmpp? can you tell me a little more?
14:33 edrin i really like subcellular localisation
14:33 edrin not right now, have a experiment running. maybe in 30 minutes?
14:33 MarkFiers fine.
14:34 edrin ok back then
14:34 MarkFiers thanks :)
14:34 egonw_ just found this in my blogroll: http://langel.files.wordpress.c​om/2008/02/real_programmers.png
14:35 MarkFiers :)
14:50 edrin MarkFiers: i am not the inventor of xmpp protocol if you mean this
14:52 edrin xmpp is a protocol currently (officially) mainly used for Instant Messenging.. in example the Google IM is based on it. AOL's AIM and ICQ just started an inofficial implementation of it - see channel title
14:53 edrin however behind the scence much more is done with it.
14:55 MarkFiers ok
14:55 edrin from a technical point of view the architecture is similar to email. a jabber ID looks like an email: edrin@jabber.org is my jabber/xmpp ID - jabber = xmpp for historical reasons
14:55 egonw_ edrin: I think MarkFiers wants to hear about your webservices architecture using XMPP
14:56 edrin ah ok
14:56 MarkFiers that was what i was searching for...
14:56 MarkFiers don't know all the terms
14:56 MarkFiers but, an xmpp based architecture is much better suited for asyncronous operation
14:56 MarkFiers that's wha i suspect, at least
14:57 edrin well, some webservices actually send the result within an email - as some services need some minutes or even houres to complete.
14:57 MarkFiers that would be another approach..
14:57 edrin with SOAP users have to ask for the result all few seconds... causing traffic.
14:58 edrin however with email it is the same. the client has to check for new mails all few seconds - causing traffic and CPU usage
14:59 edrin in xmpp the client stays connected to his xmpp-server and the xmpp-server stays connected to the xmpp-server the service is located...
15:00 edrin that means the xmpp web service can notify the client in almost realtime: "result is available now"
15:00 MarkFiers that sounds good.
15:00 MarkFiers in what state is this? is there a working prototype?
15:01 edrin xmpp is an extensible protocol. new official protocol additions are reviewd by a board that say: ok/not ok resulting in standards.
15:01 MarkFiers I think egon told me something about this
15:02 MarkFiers there is no accepted protocol addition right now? is there?
15:03 edrin i made use of existing XMPP Protocol Exstensions that almost do everything we need, added some additional specifications (asynchronous client notification for  "result is available now") and wrote a simple to use client and service implementation in java
15:04 edrin the thing we wanted to *add* to the existing protocol extensions was not clear to the people who decide
15:05 MarkFiers that is a pity... but, i understand correclty that only a facility for warning the client that the service was ready was added?
15:06 edrin i have the feeling some just read "bioinformatics" and gave up. others might have feared that the protocol does something their company is doing secretly in a similar way but earn money with it... but they have no chance if i succeed to explain the thing to the others.
15:07 MarkFiers hmm... that sounds bad.. .but did you get any founded arguments for the rejection?
15:12 edrin MarkFiers: two things would be added: one is the asynchronous client notification. in principle this is already supported by the existing protocols but nobody wrote a document "hey you can gain something (asynchronous web services) if we do things this way with existing protocols". indeed this is something that needs no authentification anyway because it is in principle supported already.
15:12 edrin However the thing we really should add is a xml based data-container that stores input and output objects. this is what the people in the board did not understand. why do we need data containers for other things (like a DNA sequence) than formulars (like html form) elements.
15:13 MarkFiers I agree...this seems imporant
15:13 MarkFiers Is there any chance of resubmitting?
15:13 MarkFiers It would be interesting to get a few services up and running, and see how it performs, compared with soap/http. Maybe it would even be possible to get biomoby running over htis..
15:15 jonalv edrin: What sort of data container do the specification already have if it's not xml based?
15:15 edrin i mean, currently the protocol is used only to transfere input output formulars for autogenerating GUI from it. that's nice. but by just adding yet another form container that can transfere other data, too... much would be gained. we called it input output object.
15:15 edrin jonalv: everything is xml based.
15:15 jonalv yes I thought so...
15:15 edrin MarkFiers: sure we can resubmit it.
15:17 edrin in this case i would make some more advertisement first. and for this i would need an existing sample implementation so people can "see something". bioinformatics workflows are maybe too abstract for them
15:17 MarkFiers you mean, the term bioinformatics, or even a bioinformatics live example would be to abstract??
15:19 edrin MarkFiers: the protocol is really great. in principle everything is already in. one limitation is transfere of big files through the single tcp stream. but even for file transferes there exist extensions. and all current servers even support to function as a file transfere proxy if people are behind firewalls...
15:19 MarkFiers what is the limit on data transport?
15:20 MarkFiers can I read up on the protocol somewhere?
15:24 edrin MarkFiers: the xmpp session is a tcp/ip stream. if you push 1GB of data through this stream you the GB will first go from the client@server1.de to server1.de to server2.org to webservice.server2.org. and all within one packet.
15:24 edrin at least the client can send only one packet after the other via the tcp stream and he has to wait for completion of packet transfer before he can send the next packet.
15:24 edrin Therefore i would recommend to see XMPP more like a command and control protocol.
15:25 edrin and the service implementations should therefore also make use of the p2p file transfere initiation protocol extension of XMPP or use yet another bioinof protocol
15:25 MarkFiers that appears the way http based webservices are going to do it to... I'm working on a metabolomics worklfow right now, and the only way i think this to be feasible is by sending uri's around, instead of thedatasets
15:26 edrin MarkFiers: but in http the client does not get informed asynchronously :)
15:26 MarkFiers that is a drawback... i agree...
15:27 MarkFiers but, for now, my only alternative... or maybe i'm missing more.
15:27 edrin MarkFiers: in addition to that xmpp has a service discovery protocol
15:27 * egonw_ is relocating to his desk at PRI
15:27 MarkFiers how does that work??
15:27 egonw_ bye
15:27 egonw_ umm... bike
15:27 egonw_ :)
15:27 MarkFiers was refering to the service.discovery.. not you relocating :)
15:27 edrin heh
15:28 edrin MarkFiers: all clients/servers/services support disco (discovery). in disco details about the service are provided.
15:28 MarkFiers are you familiar with biomoby?
15:28 edrin MarkFiers: i knwo that it is doing something like this, too
15:29 MarkFiers i was wondering how advanced the xmpp service discovery protocol is, a nice feature of biomoby is that you have a datatype (in principle, that is) and you can find all services which take that datatype as input
15:29 edrin you wan to see a javascript i wrote for testing my client implementation with my service?
15:29 MarkFiers yes
15:30 MarkFiers that is xmpp, righ?
15:30 edrin just a second
15:30 edrin yes
15:30 MarkFiers k
15:32 edrin the client is a java implementation, but as it is much more easy to test stuff without compiling java code for testing each time, i perform my tests with javascripts from within bioclipse ...
15:33 olas I really like that idea
15:34 edrin MarkFiers: if you see Sync or Async in the code this has nothing todo with asynchronous client notification. this just means if the function will block until result is available or if the function just sets a listener that will be called when stuff is complete...
15:34 MarkFiers ok
15:36 edrin do you have a pastebin link? i am currently using an xmpp client and it's complicate for me to frequently join other channels in irc ;)
15:36 MarkFiers no.. don't know pastebin
15:37 edrin moment
15:38 edrin back in a second
15:38 MarkFiers k
15:42 edrin olas, this is probably more interesting for you. this script loads my client lib in bioclipse and creates a client: http://phpfi.com/293758
15:42 edrin MarkFiers: the IExecutionPipe is just important to have nice GUI behaviour withino Eclipse IDE.
15:43 edrin it is just optional...
15:44 edrin MarkFiers: to connect you just do:http://phpfi.com/293760
15:44 edrin now you are connected to the xmpp world.
15:46 MarkFiers hmm... that was easy
15:46 egonw_ joined #bioclipse
15:46 MarkFiers it's not a very long script
15:47 MarkFiers edrin: unfortunately, i have to leave now... i'd like to continue the discussion, but that will probably not be before monday....have a nice weekend! Cheers Mark
15:47 edrin to discovere if an xmpp identy is supporting an XMPP Web Service (the service protocol i did) you can discovere the xmpp world and check if the xmpp identity is an instance of "Service" http://phpfi.com/293761
15:47 edrin ok, have a nice weekend
15:48 MarkFiers bye
15:48 MarkFiers left #bioclipse
15:49 edrin olas: what do you think of this script?
15:54 olas really cool!
15:54 olas how much with xmpp is working currently?
15:54 olas can we set up our local server here to host functionality?
15:55 olas I'd love to get started with that...
15:58 edrin i will have to fix/change some lines here and there but in principle al,ost everything is working. i can do this in two weeks (busy now), i hopefully will have more time then. but this will direct result in implementing really working test services...
16:00 egonw_ edrin: nice piece of code
16:01 egonw_ how does the callBack work?
16:01 egonw_ that is...
16:01 edrin egonw_: but in principle for GUI coding the asynchronous calls are better ;)
16:01 egonw_ when does the code return... it seems to wait now, or not?
16:01 edrin second
16:01 edrin yes it waits for 20000 ms
16:02 egonw_ but it can wait indefinately too, I assume?
16:02 egonw_ and push it into the ProgressBar ?
16:03 edrin egonw_: for this someone would have to use the following: http://phpfi.com/293770
16:03 edrin there you create a listener. if you want you can link this listener into the progress bar ;)
16:04 edrin this is something really cool  - compared to coding style of axis1...
16:05 egonw_ excellent!
16:06 edrin for scripting the Sync functions might be more easy to use though. and very nice for scripting: all xmpp identities have a user friendly implementation of .toString()...
16:06 edrin the console will show usefull info therefor
16:06 edrin good for the javascript console...
16:07 egonw_ edrin: this listener is only for finding services, right?
16:08 egonw_ not a listener that gets called when the service is done calculating, or... ?
16:08 edrin no, also for executing...
16:08 edrin here is a script that runs a service with a listener: http://phpfi.com/293773
16:08 edrin but sync is there, too ;)
16:08 egonw_ :)
16:08 egonw_ you're way ahead of me :)
16:08 edrin http://phpfi.com/293774
16:08 egonw_ these scripts can currently be run?
16:09 edrin yup
16:09 olas cool
16:09 egonw_ very cool indeed
16:09 egonw_ edrin: make sure to show these scripts to mark too, next week
16:09 egonw_ he might get interested to set up a service provider too...
16:09 egonw_ so would I, but got no time :(
16:10 egonw_ and then... with three service providers and Bioclipse as client...
16:10 egonw_ edrin: that sounds like a first author paper for edrin
16:10 edrin you see, the second Sync function is much easier to program within a script (only one line) however you can not stop/cancel the service or do a status request as you could do with the process object the Async function returns...
16:11 edrin :) yeah
16:11 egonw_ seriously... I don't think it will be much trouble getting this published in BMC Bioinfo
16:12 edrin i would really love it :]
16:14 egonw_ I'd love to help you with getting that paper out...
16:14 edrin yes, you are welcome
16:14 olas edrin: could you write up some docs on the Bioclispe wiki how to become a serviec provider?
16:14 olas I'll help as well of course (as much as I can)
16:14 edrin yes, sure
16:14 egonw_ yeah, that's what now needs to be done...
16:14 egonw_ set up some really time consuming jobs...
16:15 olas We have some here
16:15 olas we run them on clusters currently...
16:15 egonw_ yeah, like BC2 developement :)
16:15 olas some heavy data analysis
16:15 olas :-)
16:15 edrin i will finish the client implementation in the next three weeks. can spend too much time right now because i need to finish things, too.
16:15 edrin things I have to do for work
16:15 egonw_ :)
16:16 olas no worries
16:16 egonw_ yeah, the common problem
16:16 olas we all have much to do
16:16 edrin will be back later
17:13 de_maas Egon: Oeps missed a lot
17:13 egonw_ olas: CDK is very close to its 10.000th commit
17:14 de_maas You wanted me to add to possible features?
17:17 edrin me goes home, cu
17:17 edrin left #bioclipse
17:19 egonw_ de_maas: yes, please have a look at this page... comments most welcome
17:20 * egonw_ is looking up the page...
17:20 egonw_ http://wiki.bioclipse.net/index​.php?title=Bioclipse_databases
17:25 de_maas Very important issue. SD-files are the major format in our business but a lot of the biotech companies do not have software to look at them properly. If this was included in bioclipse I would definitly direct them to it.
17:26 de_maas It would also be interesting to create a local and a server (oracle) version !
17:27 de_maas MDL made a big mistake by not supporting a local version in their new client software.
17:27 de_maas They are currently fixing that.
17:27 egonw_ ok, got to run now to catch my train...
17:27 egonw_ olas: please make note of de_maas' comments...
17:27 egonw_ bye all, have a nice weekend!
17:27 de_maas Good luck on the 10000
17:28 de_maas Cheers everybody, I'm on the run as well
17:29 de_maas left #bioclipse
19:09 edrin joined #bioclipse

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