> > > Solaris find calls stat(2) on all files to
> determine
> > > whether they are
> > > directories; this requires *all* inodes to be
> > > brought into memory.
> > > 
> > > GNU find does not call stat on any of the "leaf"
> > > files because it believes 
> > > that when a directory has only 2 hard links to
> it,
> > > no entries in the 
> > > directory other than "." or ".." will be
> directories
> > > so requiring "stat"
> > > is not necessary to find more sub directories.
> >
> > Right, so this is the supposed bloat I was told
> about
> > in solaris libraries?!
> 
> I am not sure what you understan by this....
> Try to use GNU find on a mounted CD and see that it
> does
> not find anything that is not in the root directory.

I was quite surprised about what was called 'bloat' in
'Slowlaris 8'.

> 
> > > Unfortunately, GNU find's optimization is
> incorrect
> > > for a number of 
> > > filesystems and GNU find therefor may give an
> > > incorrect answer but will
> > > give it more quickly.  Solaris will give the
> correct
> > > answer in all cases, 
> > > but not as fast as GNU find.
> >
> > over 24 hours with no result versus seconds? That
> > command that was used will actually not run into
> any
> > directories. queue/mess/* represents all the
> > directories find has to go look in.
> 
> It gives typically a speedup of 5x 

I assure you that I feel far more differently about
this. This is not a mere 5x speedup. 24 hours versus
say one minute is already way beyond 5x.

Send instant messages to your online friends http://uk.messenger.yahoo.com 
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to