[MOBY-dev] Error handling in Taverna (was Re: [MOBY-l] updatingMOBY Central with input/outputarticleNames)

Thu May 18 17:56:39 UTC 2006

Hi Eddie,

Wow, this is great! Especially the secondary parameter support rocks.  
Previously I used Taverna for simple tests and for creating nice  
flowcharts of my workflows, but for the real work I used custom Perl  
clients. Now that secondaries are supported as well, I guess those  
custom Perl clients will start catching dust...

I tried a lot and so far I can not find bugs for the handling of  
Secondaries, serviceNotes and the "Parse moby data - updated (2006)"  
Local Java widget :). Just a few comments, see further down.

The "other" BioMOBY Object parser that is available form the "Moby  
Service Details" contextual menu doesn't work very well though. I  
tried to dissect an object with two childs twice. In one case the  
processor is correctly added to the workflow. Hence the output from  
the corresponding BioMOBY service goes into this BioMOBY parser  
processor, but there is only one additional output port (for one of  
the child objects). For the other child object there's no output  
port. I ran the workflow this way and I do get the correct output for  
the cild for which there was an output port. In the second case the  
parser processor was not connected to the output of the service and  
there were no output ports for the two child objects. If I decompose  
objects without children the parser works perfectly, but in that case  
one might just as well use the Local Java widget. I guess I cheered a  
bit prematurely for this one and it is a bit more work in progress,  
but it looks very promising.

On 16-May-2006, at 3:57 AM, Edward Kawas wrote:

> Hi Pieter,
>
> I uploaded a new build of Taverna that hopefully addresses the  
> servicenote
> absences.
>
> Please try it and let me know how it goes. All ports should have  
> servicenotes if
> the service provides them.

Works as advertised :). The services I use only provide "human  
readable" stuff in the notes section of the serviceNotes. If a  
service has multiple outputs then the same human readable notes are  
shown multiple times. That is a bit redundant. Unless the BioMOBY  
plugin handles the latest error handling specs and shows errors  
specifically for a certain output, maybe it would be better the have  
an additional output port for the serviceNotes instead. But for now  
I'm just very happy to see those notes.

> The build is available at
> http://bioinfo.icapture.ubc.ca/ekawas/jars/taverna-workbench-1.3.2- 
> RC1.zip

(For those who want to tried this on Linux / Unix machines: make sure  
to "dos2unix" some of the flat text files in that build.)

> As for secondarys, if you context click services that are  
> configurable from the
> 'advanced workflow ...' and choose 'Configure Moby Service' you  
> will be able to
> use secodaries.

Works! And even the enumeration lists and default values are  
supported. There is one thing missing though: I couldn't fetch  
descriptions for the secondaries. For the secondaries of my own  
services I know what they are doing, but for a remote service I  
wouldn't have a clue on how to use them. One final comment about the  
name for the contextual menus. Now there is:
* Moby Service Details: for info about the required input or output  
Simples and Collections.
* Configure Moby Service: for the optional Secondaries.
The names of the contextual menus are rather non-descriptive. Why  
would Secondaries not be part of Moby Service Details? From the  
contextual menu names I can not figure out that one of them is for  
required stuff and the other for optional parameters.

So it could be a bit more user friendly, but the bottom line is that  
I have been playing with Taverna for 1,5 day now like a little kid  
that just got an invaluable Christmas present. (Imagine what happens  
here when the object decomposition parser works as well :)).

> I was pleasantly surprised that you had found this build of Taverna  
> on the web.
> I had originally put it there for mark to download and try out  
> before I
> committed my changes to the cvs. I did think of asking you to try  
> it, but I
> didn't want to bother you.

I wouldn't mind of you'd bother me a bit more often with cool tools  
like this!

Thanks,

Pi

> Thanks,
>
> Eddie
>
>> -----Original Message-----
>> From: moby-dev-bounces at lists.open-bio.org
>> [mailto:moby-dev-bounces at lists.open-bio.org] On Behalf Of
>> Pieter Neerincx
>> Sent: Monday, May 15, 2006 5:18 AM
>> To: Core developer announcements
>> Subject: [MOBY-dev] Error handling in Taverna (was Re:
>> [MOBY-l] updatingMOBY Central with input/outputarticleNames)
>>
>> Hi Eddie,
>>
>> I tried the latest Taverna version today and I really enjoyed it :).
>> Automatic addition of articleNames works perfectly and the
>> new "Parse moby data" processor makes decomposing BioMOBY
>> objects a breeze.
>>
>> I also browsed the online documentation again and noticed
>> that the "output" and "input" ports are depreciated and only
>> meant to be used for backwards compatibility. I used them to
>> get the full output from BioMOBY services. This allowed me to
>> have a look at the serviceNotes section in addition to the
>> mobyData section. In Taverna 1.3.2RC1 (with updated *.jar
>> from the BioMOBY site) this behaviour has changed and the
>> legacy input and output ports only return the mobyData block.
>> I realise now that I've been abusing those legacy ports for
>> something they were never made for, but it was the only way
>> to have a peek at diagnostic error massages from the
>> serviceNotes. (Or is there another
>> way?) I'd love to see that kind of functionality return one day...
>>
>> Thanks,
>>
>> Pi
>>
>>
>> On 2-May-2006, at 6:51 PM, Pieter Neerincx wrote:
>>
>>> Hi Eddie,
>>>
>>> Super, I'll update.
>>>
>>> Thanks,
>>>
>>> Pi
>>>
>>> On 2-May-2006, at 3:16 PM, Edward Kawas wrote:
>>>
>>>> Hi Pieter,
>>>>
>>>> If you update your version of Taverna, you no longer have
>> to fill in
>>>> the article name as it is filled in automatically.
>>>>
>>>> Eddie
>>>>
>>>>
>>>>> -----Original Message-----
>>>>> From: moby-dev-bounces at lists.open-bio.org
>>>>> [mailto:moby-dev-bounces at lists.open-bio.org] On Behalf Of Pieter
>>>>> Neerincx
>>>>> Sent: Tuesday, May 02, 2006 2:56 AM
>>>>> To: mobydev
>>>>> Subject: Re: [MOBY-dev] [MOBY-l] updating MOBY Central with
>>>>> input/outputarticleNames
>>>>>
>>>>> Hi all,
>>>>>
>>>>> On 25-Apr-2006, at 4:53 PM, Mark Wilkinson wrote:
>>>>>
>>>>>> Hi all,
>>>>>>
>>>>>> A few months ago the spec for service registration was
>>>>> tightened such
>>>>>> that all inputs and outputs require an articleName.  This is now
>>>>>> required also in Taverna workflows, and as such, many
>>>>> "illegitimate"
>>>>>> services currently in the registry will not function with
>>>>> Taverna even
>>>>>> though they are perfectly functional in reality.
>>>>>>
>>>>>> Given that these services are not looking for an
>>>>> articleName, it will
>>>>>> do no harm to them if one were added.  Thus the fastest way
>>>>> to bring
>>>>>> all service registrations back into compliance with the API
>>>>> would be
>>>>>> for me to manipulate the MOBY Central database directly and add
>>>>>> "dummy"
>>>>>> articleNames
>>>>>> ("input", "output") to all inputs and outputs that are
>>>>> currently un-
>>>>>> named.
>>>>>>
>>>>>> Would this bother anyone?
>>>>>
>>>>> It's fine with me. Actually I already registered most of
>> my services
>>>>> to use "input" and "output" as dummy articleNames to make
>> sure they
>>>>> worked in Taverna :). I was wondering the following though. Maybe
>>>>> this is a question for Eddie. If
>>>>> * articleName attributes are required for input and
>> output articles
>>>>> and
>>>>> * every service is registered with certain articleNames for their
>>>>> inputs and outputs,
>>>>> * wouldn't it be possible to let Taverna fill in those
>> articleName
>>>>> attributes automatically?
>>>>>
>>>>> Currently I have to fill them in manually all the time
>> and that is a
>>>>> bit boring :(. Furthermore I already spent quite a bit of time
>>>>> staring at XML wondering why something didn't work ... and in the
>>>>> end it was a missing articleName or one with a typo.
>>>>>
>>>>> Cheers,
>>>>>
>>>>> Pi
>>>>>
>>>>>> Please let me know ASAP.
>>>>>>
>>>>>> Mark
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> moby-l mailing list
>>>>>> moby-l at lists.open-bio.org
>>>>>> http://lists.open-bio.org/mailman/listinfo/moby-l
>>>>>
>>>>>
>>>>> Wageningen University and Research centre (WUR) Laboratory of
>>>>> Bioinformatics Transitorium (building 312) room 1034 Dreijenlaan 3
>>>>> 6703 HA Wageningen
>>>>> The Netherlands
>>>>> phone: 0317-483 060
>>>>> fax: 0317-483 584
>>>>> mobile: 06-143 66 783
>>>>> pieter.neerincx at wur.nl
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> MOBY-dev mailing list
>>>>> MOBY-dev at lists.open-bio.org
>>>>> http://lists.open-bio.org/mailman/listinfo/moby-dev
>>>>
>>>> _______________________________________________
>>>> MOBY-dev mailing list
>>>> MOBY-dev at lists.open-bio.org
>>>> http://lists.open-bio.org/mailman/listinfo/moby-dev
>>>
>>>
>>> Wageningen University and Research centre (WUR) Laboratory of
>>> Bioinformatics Transitorium (building 312) room 1034 Dreijenlaan 3
>>> 6703 HA Wageningen
>>> The Netherlands
>>> phone: 0317-483 060
>>> fax: 0317-483 584
>>> mobile: 06-143 66 783
>>> pieter.neerincx at wur.nl
>>>
>>>
>>>
>>> _______________________________________________
>>> MOBY-dev mailing list
>>> MOBY-dev at lists.open-bio.org
>>> http://lists.open-bio.org/mailman/listinfo/moby-dev
>>
>>
>> Wageningen University and Research centre (WUR) Laboratory of
>> Bioinformatics Transitorium (building 312) room 1034 Dreijenlaan 3
>> 6703 HA Wageningen
>> The Netherlands
>> phone: 0317-483 060
>> fax: 0317-483 584
>> mobile: 06-143 66 783
>> pieter.neerincx at wur.nl
>>
>>
>>
>> _______________________________________________
>> MOBY-dev mailing list
>> MOBY-dev at lists.open-bio.org
>> http://lists.open-bio.org/mailman/listinfo/moby-dev
>

Wageningen University and Research centre (WUR)
Laboratory of Bioinformatics
Transitorium (building 312) room 1034
Dreijenlaan 3
6703 HA Wageningen
The Netherlands
phone: 0317-483 060
fax: 0317-483 584
mobile: 06-143 66 783
pieter.neerincx at wur.nl