Adrian Chadd wrote:
>
> On Fri, Feb 09, 2001, Henrik Nordstrom wrote:
> > I fully agree. So the question is if should we move that -1 to a
> > cache_dir option instead of being one of the mandatory arguments to
> > cache_dir?
> >
> > If so, we should do it before 2.4.STABLE1 is rolled, so we need to
> > decide NOW.
>
> I'm all for it.
Ok. I have now implemented a genric cache_dir options parser using the
syntax
cache_dir type path [fs-specific arguments] [options...]
fs-specific arguments are the arguments that must always be specified.
For ufs,aufs,diskd this is
size L1 L2
Global options:
read-only
max-size
diskd options:
Q1=n (defaults to 64 if not specified)
Q2=n (defaults to 72 if not specified)
Ideally I'd like to make L1 and L2 options as well, but this will break
syntax compability with Squid-2.3 so I have not yet done it.
The changes are currently residing in the SourceForge branch "fsopts",
but I will commit it to HEAD as soon as testing has finished, and unless
someone objects to the idea I'd merge the changes back to Squid-2.4.
Note: This change will in by opiniton bring the 2.4-PRE changes over the
threshold that requires yet another PRE release before STABLE1, but I
think it is worth it if nothing else because the less number of confused
users in the upgrade process..
/Henrik
Received on Sat Feb 10 2001 - 07:59:12 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:13:29 MST