[Gnu-arch-users] False "binary" files

[Sergio Callegari]

Hello,

I'm wondering whether it could be possibe to add support in arch for the 
following:
It happens quite often to have in a tree files that appear to be 
"binary" but are actually the simple result of a compression applied to 
text files or to collection of text files.
For instance I'm thinking of documentation written in the openoffice 
format (actually a zipped collection of xml files).  With the growth of 
compressed xml and the oasis open document format the number of 
situations where these "false binary files" are met is likely to grow 
dramatically.
Deltas for these files tend to be huge, even if the originating 
uncompressed files change just by a single character.
I wonder if it might be possible to make arch/bazaar aware of this kind 
of files, in order to have efficient diffs.

For instance, gz files could be expanded before diffing/taking delta.  
Similarly tar.gz or zip files could be expanded and then processed 
recursively.

Sergio

--
----------------------------------------------------------------
Dr. Sergio Callegari                 Via Fontanelle 40
Researcher                           47100 Forlì
II School of Engineering             Tel. +39.0543.786927
University of Bologna                Fax. +39.0543.786926

Affiliated with:
DEIS  - Dept. of Electronics, Computer Sciences and Systems,
       University of Bologna (www.deis.unibo.it)
ARCES - Advanced Research Center on Electronic Systems for
       Information and Communication Technologies
       Unversity of Bologna  (www.arces.unibo.it)
================================================================