[Pipet Devel] Re: Paos project

[J.W. Bizzaro]

Carlos Maltzahn wrote:
> The LOCI project looks very interesting. I'd love to spend some time on
> this. Unfortunately, I'm currently in a very busy phase of graduating so
> my contributions might be pretty small until May. But I would like to
> join.

That's no problem.  Several of the Loci developers are very busy with their real
lives right now.  I for one am trying to prepare for my second year Ph.D.
exams.  Instead of having a couple people try to do everything, we are inviting
many people to contribute, with the philosophy that many hands make light work.

> Paos has been dormant for years and I would like to revive it and make it
> more usable. Bobo is probably more sophisticated and efficient for
> retrieval but I don't know whether it supports a notification service as
> Paos does. The last time I looked at Bobo (which is also years ago), it
> was entirely web based, i.e. its front end is a web server. Paos provides
> a Client module that makes it very easy to write clients that have
> persistent connections to the Paos server, using a Paos specific protocol.

What I do not like about Bobo is that it is very much HTML-centric, as you
mentioned, and I think much of what Bobo has to offer, we don't need.  We will
be using XML and some special protocols, which we haven't nailed down yet.  I
think Paos will fit better with what we are trying to do.

By the way, the Paos Web page says it works with Python 1.4.  Has it been tested
with Python 1.5?  Will it need many modifications?

> I really like the idea of the Glyphic Command Language. For the
> Chautauqua system I wrote a graphical editor that lets you edit a
> bi-partite control graph (similar to Petri Nets). Because the editor is a
> Paos client, it also lets you observe its execution. So one idea would be
> to build a similar editor (Python/Tkinter) to construct GCL structures and
> then watch their execution.
> 

I have a few more links to systems similar to GCL:

  Clementine data miner
    http://www.isldsi.com/_borders/Image53.gif

  Lego Mindstorms RCX language
    http://www.legomindstorms.com/program/tips_tricks/tips_orgprog.html

  Lego Mindstorms Robolab
    http://www.lego.com/dacta/robolab/rcxprograms.htm

  Crickit Logoblocks
   
http://lcs.www.media.mit.edu/people/fredm/projects/cricket/logoblocks/index.html

I plan to have GCL be a part of the Loci workspace, which will consist of a
laboratory bechtop (this is where we will have objects represented as glyphs)
and a laboratory notebook (this will be a simple HTML browser for viewing
persistent analysis logs).  Oh, and we are using Python/GTK rather than
Tkinter.  Here is a page I have describing "PyG Tools":

    http://www.uml.edu/Dept/Chem/BICGroup/PyGTools/

> Chautauqua was explicitly designed to support exception handling. One
> could imagine using similar mechanisms to support exception handling in
> GCL executions so that expensive intermediate results don't get lost if
> part of the execution fails.

That sounds nice.  I think you will appreciate how well the object distribution
model will work for biological analyses.  And I think you will enjoy working
with a type of data that very few computer scientists have worked
with...Bioinformatics deals with some very unique and I think exciting problems!

> 
> Hmm, I wish I wouldn't have to write a thesis right now ...

I wish I wouldn't have to take the second year exams right now ;-)

> 
> If you are still interested in using Paos, my first contribution to Loci
> could be to write some documentation for it. Let me know.

Sure.  I think the other developers may need to get the gist of Paos first.  We
are also developing a list of loci tools that need to be developed.  I'll put
you on the mailing list (nothing automated yet--I just redistribute what I get),
and we'll see things developing (still) over the spring.


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/
--