[Pipet Devel] and still more infrastructure things

[J.W. Bizzaro]

Justin Bradford wrote:

> Well, I'm not sure this has been entirely answered. Will the wfs handle
> all the analyses from a single centralized process? Or do you still want
> for a decentralized analysis object (pathway), where each node sends the
> object to the next node, rather than going back to the wfs each time?

My intuition and experience tell me a decentralized pathway will be less
complex, work more efficiently, and be considerably faster.

> What is the distinction between a client and server?
> Is the wfs a client and the analysis tool (gatekeeper) a server?

(sigh) I'm now trying to use your terminology, I think...

    client - process performing analysis or visualization
    server - process controlling workflow and clients

What I was referring to in past e-mails is this...

    client - local machine
    server - remote machine

You can see, mixing these up can be confusing ;-)

> > Is there really any such thing as an "XML object"?  I mean, XML is a
> > way to save structured data as a _file_.  Python objects, on the other
> > hand, are data structures in memory.  We would just be going back and
> > forth between file and object using XML.
> 
> Yes. I just tend to think of the Paos object structured like the XML file.
> I guess I was basically asking for a DOM interface.

_Structured_, but we are parsing then writing.

> > So, where do we really need XML?  Could the data just be a Python
> > object?  If we need to save the object, I think it can just be "pickled"?
> 
> Well for internal network stuff, it makes sense to just use the Python
> object. Like I said, I imagine it structured something like the XML file I
> described earlier. Also, for saving a Loci analysis locally, I would
> prefer to see it written out to an XML format. The conversion would be
> fairly simple, anyway. Rather than an obscure, semi-binary format, why not
> use an easy to read text format? It'll make it easier for non-Loci tools
> to get information from our files, too.
> 
> But you're right, there's no reason to use XML for anything but files and
> maybe drag and drop (but that's not important for now). I wasn't thinking
> about that earlier.

Thinking about this a bit more, we do need to work from a file on disk because
our data can be so large.  If a user has 15-20 GUI loci opened at once, and they
are all from DNA Polymerase PDB's and 100 kb GenBank files, and all of this is
in RAM, we'll hear Scotty in the background saying, "She can't take any more of
this Captain.  She's falling apart at the seams!"

But let's reverse the question.  If we need XML files for (1) working with large
data, (2) passing data across the Internet and to CORBA systems, and (3)
archiving data, then what do we need Paos for?

I know it was my idea to choose Paos, but I'm asking if everyone thinks it fits,
and where it fits, considering the model I've been describing.

> Can Paos support complex objects, though? Actually, can Python for that
> matter? Can I have things status.analysis[5].message in both?

I'm not sure about that structure, but from what I know of Python, it'll handle
anything any modern language can.


Jeff
-- 
J.W. Bizzaro                  Phone: 617-552-3905
Boston College                mailto:bizzaro at bc.edu
Department of Chemistry       http://www.uml.edu/Dept/Chem/Bizzaro/
--