On Fri, 20 Jun 2003, Mariusz Zynel wrote:
Mirsad Todorovac wrote:
GET /gpc-hr/h-index.html HTTP/1.0
Thank you, Mariusz, for pointing out this -- the error is mine, not Maurice's. It remained from trying it with HTTP/1.1, I suppose.
No problem. But in HTTP/1.1 URI part is the same, except Host header is required. Host specifies the domain name of the host the request is made to.
See also my comments below
OK, but this won't help me -- the simple fact is that pages were displayed OK and now it's flawed. Quite a number of people in Croatia are offended if "their" letters are not displayed correctly, and they won't even read it.
Imagine a German finding Stra¤e on his page instead of Straße or Strasse ... ;-)
Fortunatelly or not, both of us have names w/o diacritic signs, but if I some important name is shown with <superscript>1</superscript> or 3/4 instead of letters in their name, they will make conclusion that (1) I don't know HTML or (2) I don't know Croatian or (3) I don't know Pascal or, the worst (4) I have something personal against (2) or the persons involved.
FME, issues involving spelling easily turn political in this areas, and I could could get out worse because of these few realtivelly insignificant bugs than if I never did anything, for in (ex-)communism those who work nothing retire safely ;-)
So, it was not the fact that I wanted server to choose which document to send -- I will explain that a static layout has been agreed upon, and that it worked so far.
So, I guess the server needs just this small adjustment (pls), and I'll sleep peacefully once again :-P
Please see continueation bellow, but it is not connected with T directly.
domac:~/c> telnet www.gnu-pascal.de 80 Trying 132.252.79.79... Connected to Tartaglia.Dida.Physik.Uni-Essen.DE. Escape character is '^]'. GET /gpc-hr/h-index.html HTTP/1.0
You're requesting the file directly leaving no choice for server to reply with another language/charset.
Well, indeed, but please see bellow.
HTTP/1.1 200 OK Date: Fri, 20 Jun 2003 19:49:56 GMT Server: Apache/1.3.26 (Unix) Debian GNU/Linux Last-Modified: Wed, 07 May 2003 01:11:57 GMT ETag: "23b2d-1a8a-3eb85d5d" Accept-Ranges: bytes Content-Length: 6794 Connection: close Content-Type: text/html; charset=iso-8859-1 ^^^^^^^^^^^^^^^^^^
Server knows nothing about the charset setting in the file. It responds with default.
Yes, but browser "sees" and can know both -- the one in the document, and the one server gives as default charset, probably for most of the documents on the server.
If we COULD inform server about the charset of each particular document, I'll be happy to learn.
Yet, so far, IMHO, the browser does the wrong thing, accepting server default charset when document-specific charset is offered. My $0.02.
<!doctype html public "-//W3C//DTD HTML 4.0 Transitional//EN">
<html lang="hr"><head> <!-- Generated automatically by make-homepage DO NOT CHANGE THIS FILE MANUALLY! --> <title>GNU Pascal</title> <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"> ^^^^^^^^^^^^^^^^^^
If you want to use content negotiation, you should rather use Accept-Language (or Accpet-Charset, or both) header in your request. Eg.
GET /gpc/index.html HTTP/1.0 Accept-Language: hr
and if there is MultiViews option used in httpd.conf and the file index.html.hr exists, it will be returned.
I think browsers use charset from the file not from headers but it may depend on the browser.
Hope it's what you guys discuss, as I missed the begining of the thread. I'm sorry for wasting bandwidth if it's not the point.
It was not wasting the bandwidth for me, but if we slide completely off-topic, we'd better not Cc: the list :-) So far I hope we hadn't.
Thanks, Mirsad