[Bug-tar] feature proposal: redundancy against data loss by corrupted archive files

[Hauke Laging]

Hello,

I recently had problems with an (uncompressed) tar file because about 8 MB 
on the CD had turned unreadable. Thus tar (and even other programs) saw 
just the first 290 MB of a 620 MB file.

(I "solved" the problem by copying the first 290 and the last 320 MB of the 
file together to a new, smaller one.)

I suggest the feature that tar can extract all administrative information 
from a tar file (everything that is not file data plus a checksum for each 
file if that's not included today) and store that in a seperate file. That 
file should not be large, maybe 5% of the original one. In case of a 
partly destroyed file it would be possible to extract most files (and to 
determine which are correct).

I should talk to the compression guys about that but it might even be 
possible to recover damaged compressed files (gzip at all and bzip2 with 
less loss) as certain data in the original file would be known.

These files might even be used for a much faster display of the content and 
faster extraction of certain files.


Best regards,

Hauke Laging

pgpQIrKD1HDLQ.pgp
Description: PGP signature