[MOBY-dev] RE: parseMobyData

[Edward Kawas]

HI!

> -----Original Message-----
> From: Rebecca Ernst [mailto:rebecca.ernst at gsf.de]
> Sent: Thursday, November 03, 2005 6:59 AM
> To: Edward Kawas
> Cc: 'Moby Developers'
> Subject: Re: parseMobyData
> 
> Hi Eddie, Mark, Pieter, Tom and others!
> 
> The problem No.1 (Collections in parsemobydata) is solved
> resp. it
> didn't exist.
> As for every other service or scavenger Taverna decomposes
> the object
> into simples! (this is true for Taverna versions up to 1.1
> and for 1.3.
> Not for the 1.2 release though but here Eddie has a bugfix
> via CVS)
> The reason we believed it is a collection problem is that
> parsing of one
> service output failed which gives back collections.
> Now we digged into it and found that Taverna doesn't have
> a problem with
> the object being a collection: the problem was that the
> service is
> registered as giving back simples but gives back
> collections. In
> contrast to gbrowse Taverna obviously only decomposes
> collections when
> it expects them to be there.
> 
> We had a look at the second problem (primitives in moby-
> objects) and
> think that your suggestion with further ports giving
> access to the
> primitives would solve the problem. If we got it right
> this would
> replace the 'parseMobyData' scavenger completely - is that
> right?
I think so. But then again, parseMobyData also allows you to
retrieve ids, namespaces, etc. I would only give you the
primitive information.


Eddie


 
> 
> Cheers,
> Rebecca and Dirk
> 
> 
> 
> 
> 
> 
> 
> 
> Edward Kawas wrote:
> 
> >Hi Rebecca,
> >
> >I was unaware of those issues, mainly because I didn't
> touch
> >that widget and I left it as is for compatibility
reasons.
> >
> >Also, I think that I could add support for parsing
> >collections. Given a collection, return a list of
> whatever
> >that widget returns.
> >
> >Moreover, you and Pieter have been talking about
> retrieving
> >the string (and most likely integers, floats, etc) from
> the
> >objects. Pieter went as far to say that he creates
> >processors to do that for him. Do you think that it would
> be
> >worthwhile to add another port to processors that contain
> >the 'primitives' so that you can retrieve them?
> >
> >For example, imagine that you are using a processor
> blastDB
> >that performs a blast search on some database and returns
> >the blast results in xml. In Moby, this would be a data
> type
> >called text-xml (I think). Would it be useful to also
> have
> >an output port that contains only the xml text in it?
> >
> >Let me know what you think. Most likely, you have a
> better
> >idea. Tell it to me and I will see what I can do.
> >
> >Eddie
> >
> >
> >
> >>-----Original Message-----
> >>From: Rebecca Ernst [mailto:rebecca.ernst at gsf.de]
> >>Sent: Wednesday, November 02, 2005 4:33 AM
> >>To: Edward Kawas
> >>Cc: mobydev
> >>Subject: parseMobyData
> >>
> >>Hi Eddie!
> >>
> >>I had another look into Taverna to see if our workflows
> >>now run
> >>correctly but found two problems of parseMobyData.
> >>The first problem is that parseMobyData can't deal with
> >>collections (it
> >>never could) the second is that it also can't deal with
> >>the new object
> >>stucture (e.g. a string object containing a string now)
> as
> >>it only
> >>checks the first layer content and never goes into the
> >>additional string.
> >>Do you know of these problems and if so - are there any
> >>plans for fixing it?
> >>
> >>Cheers,
> >>Rebecca
> >>
> >>
> >
> >
> >
> >
> >
> 
> --
> Rebecca Ernst
> MIPS, Inst. for Bioinformatics
> GSF Research Center for Environment and Health
> Ingolstaedter Landstr. 1
> 85764 Neuherberg
> fon: +49 89 3187 3583
> email: Rebecca.Ernst at gsf.de