Re: questions concerning about the range requests

On Thu, Aug 08, 2002, Songqing Chen wrote:
> squid-2.4.
> Is that processed in 2.5 and HEAD?

Uh.

Ok. Squid-2.4 and squid-2.5 should handle range requests in the strange
way you've noted.

Ie, if a range request is made and it meets some criteria, the whole
object is downloadded and the range is returned.

If not, the range is downloaded and not cached.

in squid-HEAD there's no range caching (ie if its a range request,
the request is just marked uncacheable.)

There's no real easy solution to this unless you would like to work
on the storage manager in squid-HEAD ...

adrian

>
> On Wed, 7 Aug 2002, Adrian Chadd wrote:
>
> >
> > Which branch?
> > Are you talking about squid-2.4, squid-2.5 or squid-HEAD?
> >
> >
> >
> > Adrian
> >
> >
> > On Thu, Aug 08, 2002, Songqing Chen wrote:
> > > hi, folks,
> > > I found that the range requests(requests with Range specified at the head)
> > > are partially supported.
> > > What partially I mean here is that: follow the code, it seems
> > > the request could be processed by squid, but the reply for this
> > > kind of requests will not be saved to the local disk. (In
> > > httpCachableReply, the PARTIAL_CONTENT status will not be swappable.)
> > >
> > > So my questions about this are: what prevents from saving the partial
> > > content to the disk? what if the PARTIAL_CONTENT turns to be swappable?
> > >
> > > Is that the potential "fake hit" for the later
> > > requests since range is has nothing to do with the MD5 key? If it is,
> > > I think the content may be checked from the cached data to find it is not
> > > a REAL hit and seems not very hard to do.
> > > And I also want to know the future plan of processing range requests.
> > >
> > >
> > > Thank you.
> > >
> > >
> > >
> > > --
> > >
> > > Chen,Songqing
> > > Dept. of Computer Science
> > > College of William&Mary
> > >
> >
>
> --
>
> Chen,Songqing
> Dept. of Computer Science
> College of William&Mary
>
Received on Thu Aug 08 2002 - 00:15:03 MDT