On Sat, 12 Jun 1999, Henrik Nordstrom wrote:
> Duane Wessels wrote:
>
> > The swap file number is unrelated to the filesystem layout. There
> > is a one-to-one relationship between a StoreEntry and a swap file
> > number. Thus, a single object may not occupy multiple sfileno
> > locations.
>
> I know that. But in my opinion it should be up to the filesystem to
> determine the value of this index, or else some layouts will require an
> additional level of indirection.
I think I agree with Duane. Despite its unfortunate name, swapfileno is
essentially a StoreEntry id. The module that maintains StoreEntries has a
right to assign its own ids to the objects it maintains. Additional levels
of indirection do not scare me as they are not likely to add significant
overhead, especially if StoreEntry ids are guaranteed to be allocated in a
sequential manner.
$0.02,
Alex.
P.S. I wonder if the interface will eventually allow for
StoreEntry-less Squid? Suppose I just want to have a
memory-resident Bloom filter or two for all cached URLs, and
swap-in any other per-entry information on demand. The
rationale is to use "saved" RAM for a "hot" buffer...
Received on Tue Jul 29 2003 - 13:15:59 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:12:15 MST