Hi Alan: Thanks for your email.
The process with this large iterations is just inserting a record into a MS Access DB. I pass about 19 single inputs (single outputs coming from other processes) and I execute a SQL statement to insert a new record into the database. I have a version of my workflow that works fine in windows. When i try to run it in Linux, a build-in taverna process to execute another program based on the parameters i pass to it doesn't work. The same workflow again works fine in windows including the process to insert into the DB. What i did, i wrote my own code to execute that program. Now, the last process in the workflow which is the InsertToDB is looping or running for large number of iterations. the process i wrote works fine in windows and in Linux and produces the expected results. The whole 1.5.2 workflow doesn't work in Taverna 2 Beta. For deadlines reasons, i am sticking with 1.5.2 version for now. I followed your directions and checked its metadata and then changed the iteration to use dot product instead of cross product. I am not sure what other processes need to be changed. everything else works fine, however, issue still exist with the InsertToDB process and its running for '629145600' iterations. Any other ideas? Thanks, Jamal ----- Original Message ----- From: [email protected] To: [email protected] Sent: Friday, October 9, 2009 4:41:54 PM GMT -05:00 US/Canada Eastern Subject: taverna-users Digest, Vol 41, Issue 5 Send taverna-users mailing list submissions to [email protected] To subscribe or unsubscribe via the World Wide Web, visit https://lists.sourceforge.net/lists/listinfo/taverna-users or, via email, send a message with subject or body 'help' to [email protected] You can reach the person managing the list at [email protected] When replying, please edit your Subject line so it is more specific than "Re: Contents of taverna-users digest..." Today's Topics: 1. Re: taverna-users Digest, Vol 41, Issue 3 ([email protected]) 2. Pushing Taverna execution to server and back (Paul Fisher) 3. Re: Pushing Taverna execution to server and back (Alan R Williams) ---------------------------------------------------------------------- Message: 1 Date: Thu, 8 Oct 2009 23:26:46 -0400 (EDT) From: [email protected] Subject: Re: [Taverna-users] taverna-users Digest, Vol 41, Issue 3 To: [email protected] Message-ID: <[email protected]> Content-Type: text/plain; charset=utf-8 Hi Alan: Thanks for your email. The process with this large iterations is just inserting a record into a MS Access DB. I pass about 19 single inputs (single outputs coming from other processes) and I execute a SQL statement to insert a new record into the database. I have a version of my workflow that works fine in windows. When i try to run it in Linux, a build-in taverna process to execute another program based on the parameters i pass to it doesn't work. The same workflow again works fine in windows including the process to insert into the DB. What i did, i wrote my own code to execute that program. Now, the last process in the workflow which is the InsertToDB is looping or running for large number of iterations. the process i wrote works fine in windows and in Linux and produces the expected results. The whole 1.5.2 workflow doesn't work in Taverna 2 Beta. For deadlines reasons, i am sticking with 1.5.2 version for now. I followed your directions and checked its metadata and then changed the iteration to use dot product instead of cross product. I am not sure what other processes need to be changed. everything else works fine, however, issue still exist with the InsertToDB process and its running for '629145600' iterations. Any other ideas? Thanks, Jamal ----- Original Message ----- From: [email protected] To: [email protected] Sent: Tuesday, October 6, 2009 10:42:20 AM GMT -05:00 US/Canada Eastern Subject: taverna-users Digest, Vol 41, Issue 3 Send taverna-users mailing list submissions to [email protected] To subscribe or unsubscribe via the World Wide Web, visit https://lists.sourceforge.net/lists/listinfo/taverna-users or, via email, send a message with subject or body 'help' to [email protected] You can reach the person managing the list at [email protected] When replying, please edit your Subject line so it is more specific than "Re: Contents of taverna-users digest..." Today's Topics: 1. Re: Iterations .... (Alan Williams) 2. Where is raven (Pierre-Yves) 3. Re: Where is raven (Paul Fisher) 4. Re: Where is raven (Alan Williams) 5. Re: memory management (Simon Felix) 6. Re: memory management (Stuart Owen) ---------------------------------------------------------------------- Message: 1 Date: Tue, 06 Oct 2009 09:47:37 +0100 From: Alan Williams <[email protected]> Subject: Re: [Taverna-users] Iterations .... To: [email protected] Message-ID: <[email protected]> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Jamal Alhiyafi wrote: > Hi: Hello, > I have a process that worked fine without any issues. Now, when i > try to run it, it takes for ever and it realized in the details > column when it is running that it states: > > IterationNumber='3441' IterationTotal='629145600' Active Workers = > '1'. Wow! > and the iteration number keeps increasing. This causes memory and > issues and using hte machine resources. > > What will make the operation run for that big number of iterations? The commonest reason is that you are passing around lists in your workflow. By default when a service receives several lists on its inputs then it will do a "cross product". Each element of a list will be used with every other element of every other list. So if you feed a service three lists of length five, it will iterate 5^3 i.e. 125 times. It will also produce a list that contains 125 elements. The depth of the output list will be 3, i.e. it will actually be a list of five elements each of which is a list of five list elements that each contain five values. If you then pass that list to another service together with other lists as inputs, then you will yet again iterate for the product of the lengths of the lists. If you do that a few times then you can end up with very large numbers of iterations. Taverna can also do "dot product" where the first elements of each list are used together, then the second elements of the lists, and so on. That would (for 3 lists of length 5) cause only 5 iterations, and it will output a list of five elements i.e. of depth 1. > How do I debug this issue? It is relatively easy in Taverna 2.1 beta 2 as you can look at the expected behaviour of a service. If you go to the details of a service and look at "Predicted behavior" then it will tell you the expected depth of the values on the input and output ports of the service. If the service is predicted to output a list of a depth greater than the inputs, and you are not expecting this, then it is a danger sign. To change the behaviour you need to alter the way in which Taverna is handling lists that go into the service - this is called its iteration strategy in Taverna 1 and List handling in Taverna 2. In Taverna 1, which I think you are using, click on a service and then on the "Metadata for ..." tab. You should see a button called "Create iteration strategy" - click it. The commonest requirement is to change from cross to dot product, so click on the top-level "cross product" and then click on "Change to dot product" button. You may need to do this for several services in your workflow if you are passing around lists. I hope this helps. > Thanks, Jamal Alan ------------------------------ Message: 2 Date: Tue, 06 Oct 2009 12:51:45 +0200 From: Pierre-Yves <[email protected]> Subject: [Taverna-users] Where is raven To: [email protected] Message-ID: <[email protected]> Content-Type: text/plain Dear list, I have been playing around with the sources of Taverna today but I'm now having a small problem. Where is raven ?? The class: Log4JConfiguration.java has a bunch of: import net.sf.taverna.raven... But if I do: $ find -name "*raven*" |grep taverna I get: ./taverna-bootstrap/src/test/resources/conf/raven.properties ./integration-tests/taverna-bootstrap/src/test/resources/conf/raven.properties So either I am mistaking on what is raven (which very well could be), or I am missing something. I checked these properties files but couldn't find anything related to log, repository or spi in it. Could someone help me to find out what I am missing ? Thanks in advance for your help, Best regards, Pierre ------------------------------ Message: 3 Date: Tue, 06 Oct 2009 12:01:26 +0100 From: Paul Fisher <[email protected]> Subject: Re: [Taverna-users] Where is raven To: [email protected] Cc: taverna Hackers <[email protected]> Message-ID: <[email protected]> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Hi, Maybe this should be posted to Taverna Hackers instead? Cc'd - [email protected] Paul. Pierre-Yves wrote: > Dear list, > > I have been playing around with the sources of Taverna today but I'm now > having a small problem. > Where is raven ?? > > The class: Log4JConfiguration.java has a bunch of: > import net.sf.taverna.raven... > > But if I do: > $ find -name "*raven*" |grep taverna > I get: > ./taverna-bootstrap/src/test/resources/conf/raven.properties > ./integration-tests/taverna-bootstrap/src/test/resources/conf/raven.properties > > So either I am mistaking on what is raven (which very well could be), or > I am missing something. > > I checked these properties files but couldn't find anything related to > log, repository or spi in it. > > Could someone help me to find out what I am missing ? > > > Thanks in advance for your help, > > Best regards, > > Pierre > > > ------------------------------------------------------------------------------ > Come build with us! The BlackBerry® Developer Conference in SF, CA > is the only developer event you need to attend this year. Jumpstart your > developing skills, take BlackBerry mobile applications to market and stay > ahead of the curve. Join us from November 9-12, 2009. Register now! > http://p.sf.net/sfu/devconf > _______________________________________________ > taverna-users mailing list > [email protected] > [email protected] > Web site: http://www.taverna.org.uk > Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/ > ------------------------------ Message: 4 Date: Tue, 06 Oct 2009 12:45:30 +0100 From: Alan Williams <[email protected]> Subject: Re: [Taverna-users] Where is raven To: [email protected] Cc: taverna Hackers <[email protected]> Message-ID: <[email protected]> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Paul Fisher wrote: > Hi, > > Maybe this should be posted to Taverna Hackers instead? Yes, discussions about the sourcecode are best held in taverna-hackers. I replied to the message there. > Cc'd - [email protected] > > Paul. Alan ------------------------------ Message: 5 Date: Tue, 6 Oct 2009 13:44:22 +0000 From: Simon Felix <[email protected]> Subject: Re: [Taverna-users] memory management To: "[email protected]" <[email protected]> Message-ID: <[email protected]> Content-Type: text/plain; charset="us-ascii" Hi Anja, there are two leaks in Taverna I'm aware of: 1. When you run a workflow Taverna leaks some threads. Try running your workflow over and over again while watching thread usage in jvisualvm (comes with JDK). When I ran the workflow 20 times Taverna used >100 active threads (70 of which are daemon threads). Maybe that's related to your problem... ? 2. Raven leaks classes (--> JVM runs out of PermGen space). That is probably not a problem when using workbench however. Simon > -----Original Message----- > From: Anja Le Blanc [mailto:[email protected]] > Sent: Montag, 5. Oktober 2009 09:49 > To: [email protected] > Subject: Re: [Taverna-users] memory management > > Many thanks for your answer. > > Stian Soiland-Reyes wrote: > > When you run Taverna on Windows, and launching using the taverna.exe > > file (the default), it will use up to a minimum 300 MB, but if you > > have more free memory it will use up to 50% of the amount that is > > available - calculated when you start Taverna. > > > > If this is the case for your example I guess you have about 3 GB of > > memory installed on your machine? (with ~2.2 GB free at start up). > > Yes, the machine has 3GB memory. > > > The Java virtual machine does not typically reduce the memory usage > > once it's occupied, even if the garbage collector is run and the > > memory is freed internally inside Taverna. If the garbage collector > is > > not run earlier than more memory could be used than strictly > > necessary, if the JVM decides it's faster to delay garbage > collection. > > Yes, that's the reason why I waited for 15 minutes after closing all > workflows and results. The JVM had all opportunity to run garbage > collection. > > > That said, it is still odd that you said your Taverna has allocated > > that much memory, had you been running any particularly heavyweight > > workflows when you got that number? > > I would not think so. The memory usage was build up over time. I would > change something in a workflow - let it run - delete the run - change > the workflow and so on. My impression was that for all the runs new > memory was allocated but at a delete none was freed. > > > It could also be that the memory usage you see includes shared memory > > pages used for accessing the graphic system and system libraries - > but > > in that case this bit is not 'real' memory and is actually shared > > between all other programs on your operating system. Windows Task > > Manager typically shows the "Memory (private working set)" which I > > believe excludes such memory pages. > > > > > > If you want to reduce the maximum memory allowance for Java, you can > > edit taverna-debug.bat and modify the -Xmx300m parameter to say > > -Xmx450m if you want to assign a maximum of 450 MB. Run > > taverna-debug.bat instead of taverna.exe for this parameter to be > > picked up - if you want you can modify the Taverna startup menu item > > by right clicking and selecting Properties. > > Actually I don't want to reduce the memory used by Taverna, if it needs > it. All I want is to work for a day without having to restart Taverna. > I > wouldn't have looked at memory if a workflow which had run happily > before hadn't slowed down dramatically. (I now understand that Taverna > had reached the 50% memory limit.) > > Best regards, > Anja > > > > On Thu, Oct 1, 2009 at 15:39, Anja Le Blanc > > <[email protected]> wrote: > >> Dear Taverna Developers, > >> > >> After working with Taverna for a good part of a day, I noticed that > >> Taverna was using 1,115,872 KBytes of my memory. I closed all open > >> workflows, removed all results and waited 15 minutes. The memory > usage > >> stayed the same. Could there be a problem with freeing memory, or is > the > >> garbage collection not happening? > >> > >> I don't know whether it is already reported as bug (in a quick > search I > >> have not found it). > >> > >> Best regards, > >> Anja > >> > >> ------------------ > >> System Windows XP SP3 > >> Taverna Workbench 2.1 beta 2 > >> java 1.6.0_15 > >> ------------------ > >> > >> -- > >> Anja Le Blanc > >> Research Computing Services > >> University of Manchester > >> > >> -------------------------------------------------------------------- > ---------- > >> Come build with us! The BlackBerry® Developer Conference in SF, > CA > >> is the only developer event you need to attend this year. Jumpstart > your > >> developing skills, take BlackBerry mobile applications to market and > stay > >> ahead of the curve. Join us from November 9-12, 2009. Register > now! > >> http://p.sf.net/sfu/devconf > >> _______________________________________________ > >> taverna-users mailing list > >> [email protected] > >> [email protected] > >> Web site: http://www.taverna.org.uk > >> Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/ > >> > > > > > > > > ----------------------------------------------------------------------- > ------- > Come build with us! The BlackBerry® Developer Conference in SF, CA > is the only developer event you need to attend this year. Jumpstart > your > developing skills, take BlackBerry mobile applications to market and > stay > ahead of the curve. Join us from November 9-12, 2009. Register > now! > http://p.sf.net/sfu/devconf > _______________________________________________ > taverna-users mailing list > [email protected] > [email protected] > Web site: http://www.taverna.org.uk > Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/ ------------------------------ Message: 6 Date: Tue, 06 Oct 2009 15:42:04 +0100 From: Stuart Owen <[email protected]> Subject: Re: [Taverna-users] memory management To: [email protected] Message-ID: <[email protected]> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Simon Felix wrote: > Hi Anja, > > there are two leaks in Taverna I'm aware of: > 1. When you run a workflow Taverna leaks some threads. Try running your > workflow over and over again while watching thread usage in jvisualvm (comes > with JDK). When I ran the workflow 20 times Taverna used >100 active threads > (70 of which are daemon threads). Maybe that's related to your problem... ? > Are you referring to Taverna 1 or Taverna 2? I've done some testing with Taverna 2 and didn't find any evidence of thread leaks. (I did find some evidence of a memory leak though when using beanshells). There are a few threads held by the results view after you run a workflow, but these should disappear after you close the results. thanks, Stuart > 2. Raven leaks classes (--> JVM runs out of PermGen space). That is probably > not a problem when using workbench however. > > Simon > > >> -----Original Message----- >> From: Anja Le Blanc [mailto:[email protected]] >> Sent: Montag, 5. Oktober 2009 09:49 >> To: [email protected] >> Subject: Re: [Taverna-users] memory management >> >> Many thanks for your answer. >> >> Stian Soiland-Reyes wrote: >> >>> When you run Taverna on Windows, and launching using the taverna.exe >>> file (the default), it will use up to a minimum 300 MB, but if you >>> have more free memory it will use up to 50% of the amount that is >>> available - calculated when you start Taverna. >>> >>> If this is the case for your example I guess you have about 3 GB of >>> memory installed on your machine? (with ~2.2 GB free at start up). >>> >> Yes, the machine has 3GB memory. >> >> >>> The Java virtual machine does not typically reduce the memory usage >>> once it's occupied, even if the garbage collector is run and the >>> memory is freed internally inside Taverna. If the garbage collector >>> >> is >> >>> not run earlier than more memory could be used than strictly >>> necessary, if the JVM decides it's faster to delay garbage >>> >> collection. >> >> Yes, that's the reason why I waited for 15 minutes after closing all >> workflows and results. The JVM had all opportunity to run garbage >> collection. >> >> >>> That said, it is still odd that you said your Taverna has allocated >>> that much memory, had you been running any particularly heavyweight >>> workflows when you got that number? >>> >> I would not think so. The memory usage was build up over time. I would >> change something in a workflow - let it run - delete the run - change >> the workflow and so on. My impression was that for all the runs new >> memory was allocated but at a delete none was freed. >> >> >>> It could also be that the memory usage you see includes shared memory >>> pages used for accessing the graphic system and system libraries - >>> >> but >> >>> in that case this bit is not 'real' memory and is actually shared >>> between all other programs on your operating system. Windows Task >>> Manager typically shows the "Memory (private working set)" which I >>> believe excludes such memory pages. >>> >>> >>> If you want to reduce the maximum memory allowance for Java, you can >>> edit taverna-debug.bat and modify the -Xmx300m parameter to say >>> -Xmx450m if you want to assign a maximum of 450 MB. Run >>> taverna-debug.bat instead of taverna.exe for this parameter to be >>> picked up - if you want you can modify the Taverna startup menu item >>> by right clicking and selecting Properties. >>> >> Actually I don't want to reduce the memory used by Taverna, if it needs >> it. All I want is to work for a day without having to restart Taverna. >> I >> wouldn't have looked at memory if a workflow which had run happily >> before hadn't slowed down dramatically. (I now understand that Taverna >> had reached the 50% memory limit.) >> >> Best regards, >> Anja >> >> >> >>> On Thu, Oct 1, 2009 at 15:39, Anja Le Blanc >>> <[email protected]> wrote: >>> >>>> Dear Taverna Developers, >>>> >>>> After working with Taverna for a good part of a day, I noticed that >>>> Taverna was using 1,115,872 KBytes of my memory. I closed all open >>>> workflows, removed all results and waited 15 minutes. The memory >>>> >> usage >> >>>> stayed the same. Could there be a problem with freeing memory, or is >>>> >> the >> >>>> garbage collection not happening? >>>> >>>> I don't know whether it is already reported as bug (in a quick >>>> >> search I >> >>>> have not found it). >>>> >>>> Best regards, >>>> Anja >>>> >>>> ------------------ >>>> System Windows XP SP3 >>>> Taverna Workbench 2.1 beta 2 >>>> java 1.6.0_15 >>>> ------------------ >>>> >>>> -- >>>> Anja Le Blanc >>>> Research Computing Services >>>> University of Manchester >>>> >>>> -------------------------------------------------------------------- >>>> >> ---------- >> >>>> Come build with us! The BlackBerry® Developer Conference in SF, >>>> >> CA >> >>>> is the only developer event you need to attend this year. Jumpstart >>>> >> your >> >>>> developing skills, take BlackBerry mobile applications to market and >>>> >> stay >> >>>> ahead of the curve. Join us from November 9-12, 2009. Register >>>> >> now! >> >>>> http://p.sf.net/sfu/devconf >>>> _______________________________________________ >>>> taverna-users mailing list >>>> [email protected] >>>> [email protected] >>>> Web site: http://www.taverna.org.uk >>>> Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/ >>>> >>>> >>> >>> >> ----------------------------------------------------------------------- >> ------- >> Come build with us! The BlackBerry® Developer Conference in SF, CA >> is the only developer event you need to attend this year. Jumpstart >> your >> developing skills, take BlackBerry mobile applications to market and >> stay >> ahead of the curve. Join us from November 9-12, 2009. Register >> now! >> http://p.sf.net/sfu/devconf >> _______________________________________________ >> taverna-users mailing list >> [email protected] >> [email protected] >> Web site: http://www.taverna.org.uk >> Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/ >> > > ------------------------------------------------------------------------------ > Come build with us! The BlackBerry® Developer Conference in SF, CA > is the only developer event you need to attend this year. Jumpstart your > developing skills, take BlackBerry mobile applications to market and stay > ahead of the curve. Join us from November 9-12, 2009. Register now! > http://p.sf.net/sfu/devconf > _______________________________________________ > taverna-users mailing list > [email protected] > [email protected] > Web site: http://www.taverna.org.uk > Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/ > ------------------------------ ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf ------------------------------ _______________________________________________ taverna-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/taverna-users End of taverna-users Digest, Vol 41, Issue 3 ******************************************** ------------------------------ Message: 2 Date: Fri, 09 Oct 2009 17:07:43 +0100 From: Paul Fisher <[email protected]> Subject: [Taverna-users] Pushing Taverna execution to server and back To: taverna-users <[email protected]> Message-ID: <[email protected]> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Hello, I've been running an analysis/workflow all day, which takes absolutely ages. I've now got to leave the office and take my laptop home with me, which is running the analysis. As you can probably guess, this will break the workflow and has wasted this analysis run/my day. What I would like to be able to do is to push the workflow job from my laptop to a server without interrupting it. If it is still running, when I return on Monday, I want to get the workflow back and run it continuously on my laptop to check intermediate inputs/progress/save results locally/etc, again without interrupting it. The obvious solution is to "just run it on the server, and stop being stupid!!!"; but, I didn't know how long it will take when I started it. So, I would like to push the job to the server only if I run out of time and need to leave. Is this possible? cheers, Paul. ------------------------------ Message: 3 Date: Fri, 09 Oct 2009 21:41:37 +0100 From: Alan R Williams <[email protected]> Subject: Re: [Taverna-users] Pushing Taverna execution to server and back To: [email protected] Message-ID: <e1mwmhr-0006qt...@phoebus> > From: Paul Fisher <[email protected]> > To: taverna-users <[email protected]> > Subject: [Taverna-users] Pushing Taverna execution to server and back > Hello, Hello, > I've been running an analysis/workflow all day, which takes absolutely > ages. I've now got to leave the office and take my laptop home with me, > which is running the analysis. As you can probably guess, this will > break the workflow and has wasted this analysis run/my day. > What I would like to be able to do is to push the workflow job from my > laptop to a server without interrupting it. If it is still running, when > I return on Monday, I want to get the workflow back and run it > continuously on my laptop to check intermediate inputs/progress/save > results locally/etc, again without interrupting it. > The obvious solution is to "just run it on the server, and stop being > stupid!!!"; but, I didn't know how long it will take when I started it. > So, I would like to push the job to the server only if I run out of time > and need to leave. > Is this possible? Currently no. There are issues in Jira for pausing and resuming workflows, and also for checkpointing them and re-running from a checkpoint (possibly having edited the workflow or tweaked some parameters). Your particular problem may be a special case of wanting to have a distributed engine where computational and storage resources appear and disappear; the server resources becoming available and your laptop resources disappearing for the weekend :-) That general problem could potentially be solved when Taverna moves to being within an OSGi framework. There are distributed and dynamic OSGi systems. There's another major issue hidden in your problem - not knowing how long a workflow is liable to take. Can you think about how you could have known you would have problems? Some services, such as BioMart queries, can estimate how many results they will return. We could look at combining such information with service performance data from BioCatalogue and the predicted workflow behaviour (available in 2.1 beta). That could go some way to estimating the workflow run time. > cheers, > Paul. Alan ------------------------------ ------------------------------------------------------------------------------ Come build with us! The BlackBerry(R) Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9 - 12, 2009. Register now! http://p.sf.net/sfu/devconference ------------------------------ _______________________________________________ taverna-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/taverna-users End of taverna-users Digest, Vol 41, Issue 5 ******************************************** ------------------------------------------------------------------------------ Come build with us! The BlackBerry(R) Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9 - 12, 2009. Register now! http://p.sf.net/sfu/devconference _______________________________________________ taverna-users mailing list [email protected] [email protected] Web site: http://www.taverna.org.uk Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/
