Dan Jacobson <[EMAIL PROTECTED]> writes:
> >> Hmmm, and I seem to recall /var/cache/wwwoffle/outgoing is fetched in
> >> ls -U order.
>
> A> Yes it is fetched in this order which is why it is a FIFO, entries are
> A> fetched in the order they are added to the directory.
> I had powered up on Tuesday morning and again on Wednesday morning.
> Each day I didn't finish fetching all outgoing before hanging up.
> Here we see that it is (generally) FIFO within one day, but LIFO
> across days (reboots, wwwoffle sessions, or whatever.) Interesting.
What happened is probably that after fetching on Tuesday the outgoing
directory had some space available at the bottom. When the new
requests were made offline they would fill in from the bottom. This
would give you a FIFO within one offline session. Just by chance this
fits in the empty space so that between sessions the oldest appears
last. With only two days of data it cannot be clearly shown the order
that the requests are stored.
> All determined by how Linux manages directories I suppose.
Yes this is indeed what controls the order. For the filesystem that
you use there is approximately a FIFO behaviour, but as Paul Slootman
<[EMAIL PROTECTED]> points out:
> On Mon 27 Sep 2004, Andrew M. Bishop wrote:
> >
> > Yes it is fetched in this order which is why it is a FIFO, entries are
> > fetched in the order they are added to the directory.
>
> Note that with some types of filesystem (reiserfs comes to mind) the
> order directory entries are read do not necessarily reflect the order of
> writing.
On these filesystems there will be more of a randomisation of the
order in which pages are requested. This probably has good points and
bad points. Changing WWWOFFLE to enforce a true FIFO operation will
slow things down whatever filesystem is used.
--
Andrew.
----------------------------------------------------------------------
Andrew M. Bishop [EMAIL PROTECTED]
http://www.gedanken.demon.co.uk/
WWWOFFLE users page:
http://www.gedanken.demon.co.uk/wwwoffle/version-2.8/user.html
