Re: Storing partial responses

From: Robert Collins <robert.collins@dont-contact.us>
Date: Wed, 13 Dec 2000 12:06:38 +1100

----- Original Message -----
From: "Adrian Chadd" <adrian@creative.net.au>
To: "Joe Cooper" <joe@swelltech.com>
Cc: <squid-dev@squid-cache.org>
Sent: Tuesday, December 12, 2000 8:37 PM
Subject: Re: Storing partial responses

> On Tue, Dec 12, 2000, Joe Cooper wrote:
>
<SNIP>
>
> Well, I still want to do some more work on my new store fs code.
> The storetree code is a good idea, but it needs to be completed
> (read: turned into *the* FS layer instead of being tacked on like
> it is) and I just don't have the time.
>
Anyone got some good places for me to go read about these bits of work - the store tree and you new code - quietly?

> > Possibilities:
> >
<SNIP>
>
> If you want to support partial objects in the store, you're better off
> having it implemented as part of the FS interface and allow different
> FSes to support partial objects or not.
>

Why?

Caching partial responses is a cacheability test & upstream request logic more than a file storage issue. The file storage changes
will be minimal. If we store all the obects as per the mime type of multipart/byteranges or some similar scheme, the FS capabilities
are no longer relevant. Also the big question is where should the code to create the upstream request go? That will be very tightly
coupled with the existing content to allow sensible upstream requests.

on the other hand abstracting it all back to the FS allows for very simple store logic where partial responses aren't needed. Hmm.

Rob
Received on Tue Dec 12 2000 - 17:58:12 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:13:03 MST