10-07-2004, 12:43 AM
I tried the new version with the Canterbury corpus, as i promised.
It made a file of 1434867 bytes, that makes a 51% of the original size!!
It's curious it used LZW for all files except for the Excel sheet, where it prefered a brew of differnt methods..
A suggestion. Powers of two are used in many routines. You should precalc an array of powers of 2. This way you would avoid floating point calcs in the compression routines, and everything would be much faster.
It made a file of 1434867 bytes, that makes a 51% of the original size!!
It's curious it used LZW for all files except for the Excel sheet, where it prefered a brew of differnt methods..
A suggestion. Powers of two are used in many routines. You should precalc an array of powers of 2. This way you would avoid floating point calcs in the compression routines, and everything would be much faster.
Antoni