- From: David Woolley <david@djwhome.demon.co.uk>
- Date: Wed, 5 May 2004 08:22:34 +0100 (BST)
- To: www-style@w3.org
> This feature will be quite popular, > but each http: connection takes near 40 kbytes of machine instructions just If you really mean machine instructions, this is a one time cost that is covered by the requirements for loading the page itself. If you mean the dynamic number of instructions fetched from (cache) memory during a client HTTP access, it could be right, but it does not pollute the network bandwidth, and is small compared with the processing capacity of modern machine. If you mean that this is the protocol overhead on the network, you need a better browser. Even browsers that unthinkingly let the request headers bloat probably produce well under 2KB of request headers. Of course, for images like this, there will normally be no request headers, if the web site is properly configured, as the request will be satisfied from cache. There is a penalty, especially for site entry pages (which can be, and in a true world wide web, will often be, deep links), but designed pages are already suffering from such penalties with up to 100 images. At least with corners, I can read the page and go to the next page before the I'm given a chance to view the graphic designer's artistry (assuming that layout tables with automatic layout aren't also used). With current de facto text replacement usage, I have to wait it out before I see what the links are.
Received on Wednesday, 5 May 2004 13:10:04 UTC