226941 - HTTP headers lack "; charset=UNICODE-1-1-UTF-8"

Status: RESOLVED DUPLICATE of bug 126266

(Bugzilla :: Query/Bug List, defect)

Description

[Ulrich Windl]

User-Agent:       Mozilla/5.0 (X11; U; Linux i586; en-US; rv:1.2.1) Gecko/20021204
Build Identifier: Mozilla 1.4

show_bug.cgi emits an UTF-8 charset header for HTML output, but buglist.cgi does
not. This means that any 8-bit character in the summary looks wrong in the bug
list, but looks correct when opening the bug.

Reproducible: Always

Steps to Reproduce:
1. Create a bug with summary "problem with §123"
2. Query for the bug (buglist)
3. Open the bug

Actual Results:  
The buglist shows an odd summary, the show_bug shows the correct character

Expected Results:  
Same character coding should be advertised for any page.

Comment 1

[Christian Reis]

Are you sure your installation hasn't customized its headers manually? In 2.16,
charset emission was done in each page individually, and when not specified user
agent auto-detection kicks in. I don't think we would be specifying a charset in
one page and not in the other.

Bug 126266 involves providing facilities for administrator-set character set
encodings. Note that with bug 201816 this has only to be changed in one place.
All this applies only to the 2.17 trunk of course.

Comment 2

[Ulrich Windl]

I have not customized any headers or templates in any way. I could not even find
where the UTF-8 is specified.

Comment 3

[Ulrich Windl]

I noticed that this bug does not appear in MS-Windows/XP with Internet Explorer:
It switches to UTF-8 for the buglist, but I don't know why. Maybe it's the
default. Changing the character encoding manually to UTF-8 also fixes the
problem in Mozilla, but I still think the buglist output should provide a
charset name.

Comment 4

[Dave Miller [:justdave]]

Yes, there are no charsets being specified at all right now.  If your browser is
getting any it's autodetecting on its own and not getting anything from the server.

What you can try as a quicker hack than changing all of the files is, if you're
using Apache, add a "AddDefaultCharset utf-8" to the .htaccess file in your
Bugzilla directory.

*** This bug has been marked as a duplicate of 126266 ***

Comment 5

[Ulrich Windl]

The suggested addition "AddDefaultCharset utf-8" seems to have no effect on the
output of CGI scripts. The output starts with:
--thisrandomstring
Content-Type: text/html
Set-Cookie: LASTORDER=...
Set-Cookie: BUGLIST=...

So there's no charset information present. RFC2616 suggests taht the charset
should be in the Content-Type (e.g. Content-Type: text/html; charset=ISO-8859-4).

Comment 6

[Dave Miller [:justdave]]

The directory in question needs the FileInfo privilege on the AllowOptions line
for that directory in your httpd.conf file.  Alternately, you can try putting
the AddDefaultCharset directive in your httpd.conf file in a <Directory> block
for that directory or by itself if you don't mine it applying to the entire server.

Comment 7

[Ulrich Windl]

Proof by trying:
I replaced the "Content-Type: text/html\n" with "Content-Type: text/html;
charset=UNICODE-1-1-UTF-8\n", and the problems I had were gone. I just replaced
all occurrencies in buglist.cgi 1.169.2.12 (2.16.5). Other CGI scripts have the
same problem.