[comp.sources.d] Cautious endorsement of packdisk

karl@sugar.hackercorp.com (Karl Lehenbauer) (10/16/89)

I got pack going under standard System V/386.  This was simple matter of
getting rid of the ANSI-style defines and hacking around with the include
files.

I backed up a partition and ran packdisk on it.  I aborted packdisk several
times.  Although it would leave problems in the freelist after an abort,
it never screwed up a file.  Although you should always run fsck after
a packdisk anyway, you must certainly run it any time packdisk fails to
complete to clean up the freelist.  That is only reasonable, although
perhaps packdisk should trap kills and clean up.

One thing bad about it, it takes *forever*.  Packing a badly fragmented
60 meg partition took several hours on a 386 with 28 ms ST506 hard drives.
If someone could find a way to make it go a bunch faster, that would be
great.  Even if the risk of losing something if there is a power failure
or crash was increased, it would be nice to have as an option.

Oh yeah, it's a really, really good idea to back up before packing, even though
I haven't had any problems, and don't even think about trying to pack a
mounted filesystem, even in single user mode with nothing else running -- I 
guarantee it will screw up because of contention between automatic update of 
the superblock from the in-core copy and packdisk's assumption that it is the 
only one talking to the disk.
-- 
-- uunet!sugar!karl	"There is hopeful symbolism in the fact that 
-- 			 flags do not wave in a vacuum."  -- Arthur C. Clarke
-- Usenet access: (713) 438-5018