[kepler-dev] Integration of openModeller to Kepler (original message)
Pereira, Ricardo Scachetti
rpereira at ku.edu
Thu Aug 19 13:40:27 PDT 2004
Below I'm forwarding the original message I sent to some of you
today that Matt mentioned during our conference call.
As far as I can tell, during the conference we decided that it would
be worthwhile to try to integrate openModeller to Kepler to replace the
current GARP code that is being used by the ENM pipeline. That would be
accomplished by the Oct 15th deadline.
Details of the discussion are in the call notes
From: Pereira, Ricardo Scachetti
Sent: Thursday, August 19, 2004 7:53 AM
To: Matt Jones; Chad Berkley; Jing Tao; Spears, Rodney C; Matt Jones;
Subject: RE: [kepler-dev] seek/kepler conference call tomorrow
Matt and all,
I've been following Kepler work from a distance lately, but I think
this might be a good opportunity for me to make a suggestion.
As I announced before on the seek-dev list, some coleagues and I have
been working really hard on a complete redesign of GARP and a number of
other Ecological Niche Modeling algorithms in an Object Oriented
Framework called openModeller
We are about to release the second version of the framework (v. 0.2).
The quality of the code has been improving a lot lately and we think we
have a quite stable back-end in place.
My suggestion would be for you to take some time and evaluate the
possibility of replacing the GARP code currently being used by Kepler by
the openModeller framework.
In my view, Kepler and ENM pipelines would get a number of benefits:
1) OM has a better Object Oriented design (than the current GARP
code). It might make ENM actors be a bit more elegant. It might also
reduce development time a bit when working with those components.
2) Build system uses GNU Autotools (autoconf, automake and libtool)
which is a very popular build system used for C/C++ Unix/Linux software.
It builds and runs under Unix/Linux/Windows (Cygwin). Somebody is
working right now in little details to make OM build under Mac OS X as
well (fixing one #ifdef and one #include actually).
3) OM has bindings to Python and Java generated using SWIG (Simple
Wrapper and Interface Generator). SWIG generates all the JNI code that
exposes the C++ framework objects as Java objects automatically.
4) Better data access. A significant portion of the framework provides
data access to the ENM algorithm objects using GDAL. This can
potentially make the presampling actor obsolete (at least if ENM is
implemented using OM instead of the current GARP code, but other
pipelines might use it for some other purpose).
5) Automated tests: some portions of OM are covered with Unit tests
(mostly on GARP algorithm so far, but probably expanding). OM also have
some sort of acceptance tests that were designed to test the algorithm
results, comparing actual model output with well known results.
6) Finally, OM implements the same funcionality that the current GARP
code in Kepler does.
Again, I believe that making that change would bring some benefits to
the implementation of ENM pipelines in Kepler.
I'm particularly interested in having a decent GUI interface for the
framework and Kepler seems to be a good candidate for that. So I would
be more than happy to invest some of my time as well to help implement
the ENM pipelines in Kepler using OM.
Let me know if you are interested in considering this suggestion at
Regardless of that, I would like to join you on the teleconference, if
you permit, so I can catch up on the progress of item 1 from your agenda
(ENM Niche Modeling).
Regards to all,
From: kepler-dev-admin at ecoinformatics.org on behalf of Matt
Sent: Wed 8/18/2004 2:51 PM
To: Chad Berkley; Jing Tao; Spears, Rodney C; Matt Jones;
Bertram Ludaescher; seek-dev; kepler-dev at ecoinformatics.org
Subject: [kepler-dev] seek/kepler conference call tomorrow
We agreed in Edinburgh at the SEEK meeting to be targeting an
deadline for a semi-stable Kepler release (beta1) that contains
major components to satisfy the Ecological Niche Modeling case
in SEEK. We have made progress towards this goal, but I think
we are a
bit behind. I'd like to have a conference call tomorrow to
progress and possibly reprioritize SEEK-related kepler work.
Proposed time: 10:30 am Pacific time Aug 19, 2004
We need at least the principal SEEK contributors to Kepler
Rod, Dan, Chad, and Bertram), and others are welcome. Please
know if this time works for you. If it works generally, I will
call-in details in a followup email to those who say they want
participate. Please send me agenda items, but I would like to
the following items:
1. Ecological Niche Modeling pipeline
GIS actors status
GARP and other model actor status
presampling actors status
image display actors for GIS data -- outstanding needs?
2. EcoGrid and data access in Kepler
Metadata query facility changes needed
Data query implementation status
Mammal data access -- and metadata -- DIGIR results
Environmental data layer access -- and metadata
Timeline for completion
3. Ontology-based browsing for actors and data
Java API to sparrow status
Design and implementation in Kepler status
4. Saving kepler products to EcoGrid
Prioritization of 'put' interface work
Ilkay would also like to discuss the possibility of another
mini-hackathon at San Diego appended to the September SEEK
last one was very productive for everyone who came, and it might
timing to work together for an Oct 15 beta release.
jones at nceas.ucsb.edu
http://www.nceas.ucsb.edu/ Fax: 425-920-2439 Ph:
National Center for Ecological Analysis and Synthesis (NCEAS)
University of California Santa Barbara
Interested in ecological informatics?
kepler-dev mailing list
kepler-dev at ecoinformatics.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Kepler-dev