> Doing nifty things like seeking, mmap'ing, etc, with transparently
> compressed files is not trivial to do, at least not effeciently, which is
> why you don't see it in use everywhere. Compression should be left to user
> space apps, which can be written to deal with the problem in a more sane
There are e2fs compression patches (which I would like to see in official
linux kernel some time in the future) which work relatively well. I have,
for example, a 40 megabyte database in my harddisk which I access via
mmap-calls, and, well, it works ;). I haven't really benchmarked it, but I
wouldn't say it's slow. (Not atleast when I have saved 27 megabytes.) I
have also done something actually useful, like compressed everything under
/usr/doc, although they could've been compressed with gzip (except for the
*.ps-files).
However, the point is that compression can be done in a way that it can be
useful..
> manner. Unix makes accessing good compression fairly trivial, so I think
> it's best left to userland.
But how would you compress all your source you compile daily? Compressing
file system is IMO the way to go..
'''' '' '' '' '' '' |Erkki Seppälä, erkki.s@sci.fi | ____ |
,, ,, ,, ,, ,,, | --> http://www.sci.fi/~tes/flux/| \\ ` Sigma |
'' '' '' '' '' | | \> |
,,,, ,, ,, ,, ,, |flux/Inside Informatics ry | _/__, project |
'''' '' '' '' ''' | --> http://inside.edu.lahti.fi/ | ~~~~ |
,, ,,, ,,, ,, ,, |_________________________________|_______________|