[Patch] proxy cache for 304 Not Modified

[Woon Wai Keen]

On 2011-09-19 6:47 PM, Maxim Dounin wrote:
> Additional question to consider: what should happen if original
> 200 reply comes with "Cache-Control: max-age=<seconds>" (or
> "Expires:<time>").  Should we use it?  Or should we use another
> one from 304 reply?
>
> Please provide rationale based on RFC2616.
I'm most interested in this point you raised.

Excerpt from RFC2616 13.5.3 (snipped only the relevant parts):
>    If the status code is 304 (Not Modified), the cache uses the entity-
>    body stored in the cache entry as the entity-body of this outgoing
>    response.
>
>    The end-to-end headers stored in the cache entry are used for the
>    constructed response, except that
>
>       - any end-to-end headers provided in the 304 or 206 response MUST
>         replace the corresponding headers from the cache entry.
Due to this, we need to read the cached headers, merge it with upstream 
headers, write the updated cache to a temp file and replace the existing 
one.

To avoid the copy IO for some specific use cases where 304s won't have 
consequential header updates, it can be sufficient to just update 
valid_sec. Of course this would be highly uncompliant with RFC2616...