[Pipet Devel] Re: Express

[J.W. Bizzaro]

Hi Conrad!

Wow.  I thought maybe you died or something ;-)

Conrad Parker wrote:

> I'm planning on making Express handle XML applications nicely, though I haven't
> looked into it much yet. Insofar as my contribution involves writing a
> browser which can support various XML applications, yes I'd like to be involved
> with TULIP :) Beyond that I don't think I can help much - my knowledge of
> biochemistry doesn't extend too much beyond high school and brief encounters in
> studying information theory and genetic algorithms :)

We have 9 bioscientists on our team, so you need not worry about that.  Time has
passed, and at this point we are looking for code to a generic XML browser, but
something we can build upon.  Each GUI tool in Loci (the name Tulip is being
phased out) will be a special-purpose XML browser and will support one (probably
one) XML definition.

Also, we are working with Python/C with bindings to GTK/GNOME.  So, we need
something we can wrap some Python code around.  We do have bindings to all of
the GTK/GNOME widgets, so we may be able to make the whole thing in Python.  But
of course native C will be faster.  Which would you recommend?  I haven't looked
into the speed requirements for a browser, but what we need will be graphics
intensive.

> cool :) looking at your developer's page, if Jay Painter is working on the
> BSML implementation then it should probably be ok for me to just do the web
> browser support (which will of course give networking etc).

Unfortunately, Jay is tied up with GNOME development for RedHat and may not come
back to Loci.  He was our only GTK/GNOME expert.  So, I guess we'll have to
start with page one of the tutorial :-P

> The reason I mention it is because its architecture is similar to your ideas
> for TULIP. In particular, looking at your ideas for GCL (do you have an
> implementation yet?) it looks like the way you want to be able to connect up
> components (tools) is similar to the way aube works - however aube's system is
> currently entirely graphical (ie. you can connect up various components, but
> not load/save the state of connections). I am looking at using XML to handle
> this information, as it can save parameters of each component more cleanly than
> a scripting language could.

We are just now trying to implement an active object server for Python (Paos, by
Carlos Maltzahn).  And we're talking about making a workflow system so that XML
objects can be juggled and tracked.  I agree that XML is a nice way to handle
this sort of data.

> So, if you'd like to save yourself some coding you can use the system I've got
> going with aube, including some widgets for selecting inputs and connecting up
> components. I'll soon be adding an overview widget for editing the whole graph
> of connections

Aube looks very nice!  You know, we _really_ need a GTK guru to put the graphics
together for GCL (all the glyphs and arrows).  This may be even more helpful to
us than an XML browser, and it wouldn't require any knowledge of biology.  I
imagine you might be able to make use of GCL in Aube or other programs (in fact,
I think GCL is the way to go for user-friendly UNIX).  Is this something you'd
like to take on?  Have you tried to make dnd icons and user-manipulated graphics
with gnome-canvas?


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/
--