- From: Dave Kristol <dmk@allegra.att.com>
- Date: Fri, 16 Dec 94 16:45:29 EST
- To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Regarding new attributes for the IMG tag, to specify width and height: Keep in mind that the person/program that creates the IMG reference may not actually have an image in hand -- it may be on another server. That's one problem with the proposals to have a server bundle images along with the base document: the server may not actually have them. It's also a problem for whoever creates the base document: they may not know its size, so they can't edit the IMG tag. Oh, they could peek, but since the image is elsewhere, the owner of the image could change what the image URL refers to, and then the information in the IMG tag would be wrong. (No flames about URNs, please!) I'm nervous about solutions that would require the server to scan documents (even if they cache the result) to locate IMG requests. On the other hand, if the image information were part of the server's database (using the term loosely), the whole affair would "fail-soft". For example, with the scheme Eric Sink described, if a client didn't get response headers that describe the images in the base document, it would simply make a best effort to render the page. But if the server helped the client by providing the correct (would be nice!) size information, the client could do a nice rendering job as it receives the base document. Dave Kristol
Received on Friday, 16 December 1994 14:19:45 UTC