- From: Natchu Vishnu Priya <vishnu@cs.iitm.ernet.in>
- Date: Fri, 21 Mar 1997 16:09:16 +0530
- To: http-wg@cuckoo.hpl.hp.com
- Cc: vishnu@cs.iitm.ernet.in
Hi, I have been studing the HTTP/1.1 specs and have noted with interest the changes to HTTP/1.0 which formally define the requirements for agents, to use peristant connections and caching. There seems to be a certain incompatibility in the methods used to achieve caching the ones to reduce latency and bandwitdh requirements by the use fo persistant connections. Put simply persistant connections require me to place all my related documents, images, etc on a single comman server, while to facilitate caching it might be very useful to use a comman url for inlaid images. For example an entire university might use images located on a master image server and use a single URL for refering to their emblem. Also certain very comman images almost always exist in the cache (like the icons used by the apache web server (for mime types..), etc) Since caching still (and this is a real disappointment) depends on the URL matching exactly and not on some new unique name space both the requirements cannot be achieved concurrently. I have a couple of suggestions for a new scheme which allows for the use of the present URL scheme while still retaining the benefits of both persistant connections and caching. I don't think I have been particularly clear on the topics I have mentioned but I don't hope the main point is noted. I'll come up with a page that explains these is a better fashion, but don't you all think this is an issue that has to be opened. vishnu vishnu@cs.iitm.ernet.in
Received on Friday, 21 March 1997 02:55:18 UTC