webserver upgrade forcing charset=ISO-8859-1?

[Russ]

Something seems to have changed recently with the handling of charset for webpages served from keenspace. I suspect it might have to do with recent apache upgrade I see mentioned by Kisai at the keenspace main page.

I have utf-8 in my pages which has displayed foreign text just fine forever until recently. Check the newsbox of http://comicollage.keenspace.com/ or most any recent page of http://kofightclub.com - the Esperanto text is screwed up now.

I know enough about valid html to know

<meta http-equiv="Content-Type" content="text/html; charset=utf-8">

should make the text display fine (and indeed it does on hundreds of non-keenspace pages I've made (which validate with http://validator.w3.org), and on keenspace pages until sometime in the last few days), so I'm guessing the keenspace webserver has had something changed. I can take the actual index.html from my public directory and display it fine on another server or locally, for instance.

Further examination shows that the webserver is forcing ISO-8859-1:

russ@zachs:~$ telnet comicollage.keenspace.com 80
Trying 66.220.2.7...
Connected to keenspace.com.
Escape character is '^]'.
HEAD / HTTP/1.1
Host: comicollage.keenspace.com

HTTP/1.1 200 OK
Date: Fri, 03 Dec 2004 03:07:28 GMT
Server: Apache/2.0.52 (FreeBSD) PHP/5.0.2
Last-Modified: Fri, 03 Dec 2004 02:17:02 GMT
ETag: "be244e-6681-bbf5e380"
Accept-Ranges: bytes
Content-Length: 26241
Content-Type: text/html; charset=ISO-8859-1

And this Content-Type apparently taking precedence over the charset in the webpage itself.

From other websites (where the valid pages display correctly) this just shows Content-Type: text/html - i.e. without explicitly forcing a charset.

Can things be set back to the way they were until recently? Otherwise vast amounts of text on my kofightclub.com site are screwed up.

[Kisai]

The server is not being downgraded back to apache 1.3.x

I've removed the setting from apache 2 however that is sending that header.

Apparently according to the instructions in the conf file, the header is supposed to always be sent, so there is a possiblity that the web browsers are not correctly. Remember UTF-8 files have bytes at the beginning of the file that indicate they are UTF8.

[Russ]

The server is not being downgraded back to apache 1.3.x

I've removed the setting from apache 2 however that is sending that header.

Apparently according to the instructions in the conf file, the header is supposed to always be sent, so there is a possiblity that the web browsers are not correctly. Remember UTF-8 files have bytes at the beginning of the file that indicate they are UTF8.

I didn't mean I wanted you to downgrade the server; just retain the earlier more sane behavior with respect to not overriding the charset in the html page itself. I have verified the header is still being sent now, but without the charset override (i.e. it again just has "Content-Type: text/html" instead of "Content-Type: text/html; charset=ISO-8859-1"), so yay, everything is cool again. Problem solved. Thanks!