Robert Bihlmeyer wrote:

>>Also, HTTP/1.1 requires that any compliant server includes Date header
>>with all responses (unless the server is clockless). 
>>(http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.18 )
> 
> Yeah. But fixing that is pretty low priority for me until you show me
> that it breaks something other than a spec (e.g. a real client). We do
> include Date in /some/ responses.

The fix should be oneliner and it needs to be done anyway for proper 
caching support. One can't tell what will happen with different 
browsers/readers (excluding IE which doesn't care what kind of garbage 
you feed it with) and caching behavior if the spec is broken.

> In a related vein, DNF and friends should perhaps be 404s.

IIRC they already are..

> My current caching target are the web interface images (and static
> pages), as these make most sense to cache. For freenet content, I
> don't know, we may actualy want to actively prevent caching.

Making the 304 response work for FProxy pages should fix its caching 
issues. ETag could be useful for actual Freenet content where you can't 
exactly tell the last modified date. Every content sent to the browser 
takes cpu and ties resources in FProxy thus should be avoided if 
possible. And if you want real life example, most browsers limit 
connections (and requests with HTTP/1.1 pipelining) to proxy/hosts and 
giving fast responses frees those connections for retrieving next 
content. If a page in Freenet has lot of seperate content (images), it 
takes somewhat long time to reload the page because FProxy has to 
process and send all the content again to the browser even if it was 
already cached.



_______________________________________________
devl mailing list
devl at freenetproject.org
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to