søn, 14.12.2003 kl. 19.58 skrev Børre Gaup:
Wednesday, December 10. b. 2003 23:37, Tommy Gildseth don čállet:
Keld Jørn Simonsen wrote:
On Tue, Dec 09, 2003 at 05:03:39PM +0100, Tommy Gildseth wrote:
If the server already sends ISO-8859-1/15 as the content-type, the browser(at least Mozilla) will prefer this over what is specified in the HTML document.
Are you sure? I believe the browser will always follow the meta tags of the docyment, if present, as this is a way to override the system wide settings of the server. Would be silly the other way around (- or give less capabilities), and I believe this is not well defined in a RFC, while meta tags are well defined.
As I said, this is the behaviour I have experienced that Mozilla has. I haven't tested it with other browsers, so I couldn't really tell if the same goes for Opera and IE etc.
This is what you can expect, for it is indeed what the standards mandate. Which means setting the encoding server-wide is not such a good idea. After all, latin-1 is supposed to be assumed when no encoding is specified. For us, latin-1 is an acceptable _fallback_, but we want utf-8 on many of our pages.
I have experienced the same thing. I uploaded some utf-8 pages to d.s.n/ ~albbas/. The pages had the meta-utf-8 thingie defined, but the pages displayed the latin1 version of the utf-8 characters. I tested with mozilla 1.5 and konqueror 3.1.4.
Ask "drift" to disable the default encoding in httpd.conf. It is not needed, and causes a lot of trouble.