- From: Brian Behlendorf <brian@organic.com>
- Date: Wed, 10 May 1995 13:57:11 -0700 (PDT)
- To: Larry Masinter <masinter@parc.xerox.com>
- Cc: DLEVINE@ssf4.jsc.nasa.gov, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
On Wed, 10 May 1995, Larry Masinter wrote: > As for a client-supplied link-speed estimate, I'll just point out that > this is something that the server may be in a better position to > estimate than the client. True, but hopefully the web site designer is already compensating for that (by not using big images when they're on the slow end of a 14.4, etc). The server load and bandwidth are known entities - the client bandwidth isn't known, and is what we're looking for ways to express. The "mxb" and "mxs" parameters are close, but not quite the cleanest way to express *in*a*relative*way* whether I want high-bandwidth information or low bandwidth information. Consider the following two versions of an object the server can provide: 1) High bandwidth: a 20K HTML page (lots of formatting and style sheet info) with 2 50K inlined logo images and 10 5K inlined icons. 2) Low bandwidth: a 10K HTML page with 2 15K logos and no icons. Presume that the inlined images of one page wouldn't make sense in the other - the 15K logos *aren't* a direct replacement for the 50K logos. How do I express, in a general sense, that I'd rather see #2? I don't want to set my text/html mxb to be 15K, because there are lots of >15K documents I'd probably be happy to wait to download. And I can't vary it on image/gif, because the html page has already loaded; and even if the images were interchangeable, a provider could easily just break up their large 50K image into 5 10K images and I'd still be left with a heavily loaded page. Conceivably, my browser could have a concept of how much data per page I'm willing to accept, i.e. "I want 50K or less per page" - after downloading and measuring the 10K HTML page, it sets the mxb to be 8K for each of 5 inlined images. But even that is clumsy. I guess the main problem is that people will be thinking of "how long I'll wait for this page to render" in terms of seconds (which over a slow pretty much directly maps to bytes), and not in "how long I'll wait for each part of this page". And even then there are other factors - how the page is laid out, formatting hints, etc. Expressing that in an absolute limit on page elements doesn't seem to be the right way for the server to know, big-picture, what the client wants. Brian --=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-- brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/
Received on Wednesday, 10 May 1995 13:58:07 UTC