- From: Alan J. Flavell <flavell@a5.ph.gla.ac.uk>
- Date: Wed, 16 Dec 1998 12:00:45 +0000 (GMT)
- To: Charles McCathieNevile <charles@w3.org>
- cc: WAI GL <w3c-wai-gl@w3.org>
On Tue, 15 Dec 1998, Charles McCathieNevile wrote: > While Apache is the single most popular webserver there are many people > using other systems. Indeed; but it doesn't change the HTTP protocol. If their servers offer equivalent features by a different means, then they would need to consult their documentation for details. > However there is, in HTML, the META HTTP-EQUIV > construction, which effectively allows pages to specify their own headers. Excuse me, but this construction is far from being as "effective" as you imply. The original idea seems to have been for servers to parse the header part of the file and send out real HTTP headers based on the contents of the META, but that has not been widely deployed in practice: most of the practical support for META on the WWW has turned out to be _clients_ parsing the META headers in HTML documents and then behaving as if they had received those HTTP headers. However, the extent to which this actually works on the WWW is decidedly limited, and sometimes even counterproductive (I could give examples, but this would perhaps take the discussion off on a tangent.) And it doesn't work at all for WWW documents that aren't HTML. > This can (usually) be successfully exploited in one of the situation where > control is impossible - the use of a CD, disc, etc to provide a website. That's true, but is to my mind quite a different situation, with a different range of requirements and compromises. However, I would suggest with respect that we have now explored the various issues, and I would not want to press the case over-much. Nevertheless, I do stand by my view that there is a good case for proper use of HTTP headers. best regards
Received on Wednesday, 16 December 1998 07:00:49 UTC