OLE4BioTech

Andrew Dalke dalke at bioreason.com
Mon Dec 7 14:25:22 EST 1998


deodiaus at my-dejanews.com commented:
> I see the software industry relying heavily on Component Object
> Modeling, as successfully marketed by Microsoft. 

  COM is a de facto standard for a single proprietary platform
(although there are some 3rd party implementations of COM for
unix).  Its use constrains me from using specific hardware which
I find useful (stereo OpenGL, >4 processor boxes, fault
tolerant machines, >4GB memory, etc.) and a wide body of software
which is not designed for the Microsoft platforms.

  A potentially more useful system, CORBA, is available which
runs on a wide range of platforms.  "Open source" versions exist
for several COM servers and client implementations (ILU, CORBit),
and bridges exist between CORBA and COM for some interoperability.

  Given your statements for the GNU and Linux development models,
you may have come across references to CORBA in discussions of
GNOME, the GNU Window Manager.

> One advantage OLE has over Java is that it is a binary interface
> standard which uses fast platform specific code without having to
> be byte code interpreted at run-time.

CORBA is a much more appropriate comparison than Java (for after
all, using the COM interface you cannot write software, you only
control existing software -- or can you in COM give me 2+2 without a
COM server somewhere?).

> Basically, I would like to set up a working group similar to DMAC
> for Biotech modeling

  There exist CORBA component architecture proposals already for the
"life sciences."  Take a look at http://www.omg.org/homepages/lsr/ .

> Secondly, we need to write a freeware modular (using development
> model as set forth in the GNU copy-left agreements or as is being
> done in Linux) or modify an existing [freeware with copyleft
> disclaimer] modular which adheres to these standards.

  Take a look at Bioperl (http://bio.perl.org/) for one example
or MMTK (http://starship.skyport.net/crew/hinsen/mmtk.html) for
another under Python.

  The field of computational biology/chemistry is wide and complex
and the people few.  The end goal is worthy but it will be a hard,
long road to get done what you propose.

						Andrew Dalke
						dalke at bioreason.com




More information about the Comp-bio mailing list