From: sct@dcs.ed.ac.uk (Stephen Tweedie) Subject: Re: New feature for the filesystems. What do you think ? Date: Sat, 24 Apr 1993 20:06:00 GMT
In article <pdhC5x8Kn.1u1@netcom.com>, pdh@netcom.com (P D H) writes:
> I'll still prefer to use better compression that can take advantage
> of larger amounts of data to compare, and can squeeze data in
> tighter by not having to worry about the blocksizes.
The ext2fs compression code I am developing will byte-align all of its
compression blocks within the filesystem's blocks - no wasted space.
The only extra padding space will be to fill the entire compressed
file out to a block boundary, and to pad sub-blocks if necessary to
ensure that the block headers don't span filesystem block boundaries.
The compressed file *must* still be block structured in order to
support random read access (necessary for demand paging of executables
and shared libraries), so I won't be getting rid of this.
Once the filesystem supports fragments, the end-of-file wasted space
will become much less of a problem.
However, I'm still not sure what the best size for compression blocks
will be - I will be making this a tunable parameter, and I suspect
that 8K will probably be a good choice. Any thoughts on this?
Cheers,
Stephen Tweedie.