Estimating a code regularity proxy of the Mastrave modelling library, Octave

Language: Layout:

disp( 'Estimating a code regularity proxy of the Mastrave modelling library' )
disp( '--------------------------------------------------------------------' )

disp( '   ' )
disp( 'About the Mastrave modelling library (http://mastrave.org ):'                )
disp( '   Mastrave is a free software library written to perform vectorised '       )
disp( '   scientific computing and to be as compatible as possible with both '      )
disp( '   GNU Octave and Matlab computing frameworks, offering general purpose, '   )
disp( '   portable and freely available features for the scientific community.  '   )
disp( '   Mastrave is mostly oriented to ease complex modelling tasks such as '     )
disp( '   those typically needed within environmental models, even when '           )
disp( '   involving irregular and heterogeneous data series.'                       )

disp( '   ' )
disp( 'About code regularity:' )
disp( '   Source code complexity is measured with a great variety of metrics.'      )
disp( '   Lines of code (LOC) or the Mc-Cabe''s cyclomatic complexity (MCC) '       )
disp( '   are frequently used but also widely debated.'                             )
disp( '   As a very simple index for discussing the cognitive complexity of code, ' )
disp( '   in http://dx.doi.org/10.1145/2597008.2597140 it is suggested that '       )
disp( '   code regularity "may significantly reduce complexity, because once one '  )
disp( '   figures out the basic repeated element it is easier to understand '       )
disp( '   additional instances".'                                                   )
disp( '   Their proposed "metric for regularity is based on the observation that '  )
disp( '   regularity lies at the basis of text compression. [...] Therefore,'       )
disp( '   the compression ratio can be used as a metric for regularity.'            )

disp( '   ' )
disp( 'About code regularity in Mastrave:' )
disp( '   Mastrave implements the Semantic Array Programming paradigm and is'       )
disp( '   structured using a detailed coding style for algorithms and comments.'    )
disp( '   This disciplined coding style should generate a noticeable level of'      )
disp( '   code regularity.' )
disp( '   For testing this hypothesis, the GNU Octave/MATLAB functions of the '     )
disp( '   library have been considered. The compression ratio of single files and ' )
disp( '   of the whole set of functions is computed with GNU gzip (option -9).'     )
disp( '   The results show that part of the code regularity emerges when '          )
disp( '   considering the coding style consistency of the whole set of functions. ' )
disp( '   ' )

function  answer = mfiles_do( command )
   [ ~, answer ] = unix( [ 'mfiles=$(mastrave -i | sed -r "s,sh/.*,m,"); ' command ] );
end

uncompressed_bytes = mfiles_do( [                                                   ...
   'du -c --bytes $mfiles/*.* | tail -1 | tr "\t" " " | cut -d" " -f1'              ...
] )

disp( ' ' )
disp( 'Method 1 : simple compression as zip archive (file by file)'   )

compressed_bytes   = mfiles_do( 'gzip -9 -c $mfiles/*.* | wc --bytes' )

compression_ratio  = 1 - str2num(   compressed_bytes ) / ...
                         str2num( uncompressed_bytes )

compressed_bytes_static = mfiles_do( [                                              ...
   'tempfile=$(mktemp); gzip -9 -c $mfiles/*.* > $tempfile; '                       ...
   'cat $tempfile | wc --bytes '                                                    ...
] );
% Consistency test
assert( compressed_bytes_static == compressed_bytes )
disp( ' ' )
disp( 'Method 2 : zip compression from a tar archive (whole set of files)'         )

disp( 'Method 2a: .tar.gz directly generated with GNU tar:'                        )
compressed_bytes  = mfiles_do( 'GZIP=-9 tar -czPf - $mfiles/*.* | wc --bytes '     )
disp( 'Method 2b: .tar.gz from a .tar file (GNU tar) then compressed (GNU gzip):'  )
compressed_bytes  = mfiles_do( 'tar -cPf - $mfiles/*.* | gzip -9 -c | wc --bytes ' )

compression_ratio = 1 - str2num(   compressed_bytes ) / ...
                        str2num( uncompressed_bytes )

disp( ' ' )
disp( 'List of processed files:' )
processed_files = mfiles_do( [                                                      ...
   'tempfile=$(mktemp); ZIP=-9 tar -czPf - $mfiles/*.* > $tempfile; '               ...
   'tar -tPf $tempfile | rev | cut -d/ -f1 | rev'                                   ...
] )

[ + ] Show input
Absolute running time: 3.25 sec, cpu time: 2.97 sec, memory peak: 50 Mb, absolute service time: 3.32 sec