Timeline

OMPC has been under development by Peter Jurica at the Perceptual Dynamics Laboratory (PDL) at RIKEN Brain Science Institute roughly since 2006. Until sufficient interest had built around the application it was used only in projects with a small group of researchers at the PDL and collaborating institutes. Since the group of people was very small OMPC was never used in a fully automated fashion. Since the first public release on 15th of September 2008 it possible to translate MATLAB(R) code to Python compatible syntax without human assistance or intervention of any kind. It should take another month to catch a lot of bugs that are inevitable for any first release.

As it is explained in the documentation, the code that is generated by OMPC is fully compatible with Python syntax. This code however relies on the ompclib which implements a number of base objects that allow emulation of MATLAB®'s features. It is sad that because of legal issues it is not possible to simply translate MATLAB®'s library. This should be possible on a single user basis if a user owns a license for the code he/she is transforming. Be aware that you cannot share such code. It is possible to apply translation on Octave's library that is available under the GPL license. This could make transition towards an open MATLAB® compatible platform faster. However Octave itself is not 100% compatible with MATLAB® and those steps should be taken with extra caution.

 Public release 1 September 15th 2008

OMPC compiler made available for public testing. It is expected that a number of problematic issues will appear as testing until this point was done only by a closed source of MATLAB® users with their own style and habits of coding. MATLAB® parser is surprisingly flexible and often it is hard to guess what is the correct syntax. There is no public formal specification of MATLAB®'s grammar.

OMPClib contains all the objects that allow running OMPC generated code. A small portion of MATLAB® compatible function is available for the purpose of demostration of features.

From this release an online compiler is available at http://ompclib.appspot.com/ that allows upload of m-files submitted by outside users of OMPC.

OMPC is available for download.

 Public release 2
October 15th 2008 (I am late, but will finish releasing by the end of October)

The second release accompanied with a small portion of "OMPC standard library" that includes 2D plotting based on pylab. At this point we hope to decide the final form for the target syntax of the .pym file.


Random releases through the Mercurial repository at BitBucketDecember 2008

It turned out that some people are already using OMPC for translation and write their own ompclib based on the documentation.
The development will now focus on fixing many small issues with the parser a development of the ompclib.
Milestone 1April, maybe May 2009
or possibly July 2009

These are under development from the beginning of 2009:
 - parser restructured, cleaned and modularized
 - ompclib with basic linear algebra, FFT, sparse matrices, strings and structures (for load and save)
 - basic plotting 2D
 - some of the tutorials from http://www.mathworks.com/products/matlab/demos.html

 Milestone 2autumn of 2009

OMPClib.juricap.com, an on-line application. The intention is to make toolboxes and applications shareable and available in a wiki-style Python package.

Comments