[Biojava-dev] Fwd: [Biojava-l] Port an Algorithm to Java

[Dragos-Bogdan Sima]

---------- Forwarded message ----------
From: Dragos-Bogdan Sima <simadragosbogdan at gmail.com>
Date: Wed, Apr 4, 2012 at 2:21 AM
Subject: Re: [Biojava-l] [Biojava-dev] Port an Algorithm to Java
To: Amr AL-Hossary <amr_alhossary at hotmail.com>, Scooter Willis <
HWillis at scripps.edu>, Andreas Prlic <andreas at sdsc.edu>


Hello everyone,

As I went depth to the algorithms of Blast+ trying to figure out what
whould be the best way to aproach the porting to Java, I came up with a new
idea.

What if I first search for a source .cpp that does not have any
dependencies (it does not need for another source .cpp for its processing)
- "terminal source". For that source I could easily found out what is its
input (print it in a file) and what is its output (print it in another
file). Then in a java environment I would port that source to java and test
manually (not necessary if I implement a test function) for the saved
input, if its output is the same with the saved one. If true then the
porting should be correct.

After processing with all the "terminal sources" I would advance with
another level (directory) for each "terminal source" until all the porting
is done.

Please tell me if I am thinking correctly and this approach is practical.

Have a nice day,
Dragos.