Re: www.fvwm.org web server

2007-08-24 Thread Scott Smedley
Hi Renato,

  If so, I am just asking if it's possible for the www.fvwm.org web
  server not to specify/override the encoding. Then, presumably, the
  web browser would determine the encoding from the HTML header  it
  would render correctly.
 
 Wouldn't it be easier to change the html man page encoding instead?
 This problem is only in the man pages, so I suppose that would be an
 easier fix..

The HTML pages are automatically generated by xsltproc using the DocBook
stylesheets. So we'd have to modify the stylesheets which are effectively
a standard.

To my mind, the problem is the configuration of the web server (it's
valid to generate an ISO-8859-1 encoded web page) - we should be fixing
the _cause_ of the problem, if we can.

Again, this is not my area of expertise. If someone has a better grasp
of what's going on  what should be happening, please let me know.

Scott.



Re: www.fvwm.org web server

2007-08-24 Thread Björn Steinbrink
On 2007.08.25 11:53:04 +1000, Scott Smedley wrote:
 Hi Renato,
 
   If so, I am just asking if it's possible for the www.fvwm.org web
   server not to specify/override the encoding. Then, presumably, the
   web browser would determine the encoding from the HTML header  it
   would render correctly.
  
  Wouldn't it be easier to change the html man page encoding instead?
  This problem is only in the man pages, so I suppose that would be an
  easier fix..
 
 The HTML pages are automatically generated by xsltproc using the DocBook
 stylesheets. So we'd have to modify the stylesheets which are effectively
 a standard.
 
 To my mind, the problem is the configuration of the web server (it's
 valid to generate an ISO-8859-1 encoded web page) - we should be fixing
 the _cause_ of the problem, if we can.
 
 Again, this is not my area of expertise. If someone has a better grasp
 of what's going on  what should be happening, please let me know.

Likely (in my experience), the Apache configuration has the
AddDefaultCharset option set to UTF-8. Debian started doing that some
time ago, but reverted it by now, as it easily causes trouble unless you
know what you're doing. I guess Fedora did the same (well, except for
the revert).

That configuration setting tells Apache to append ;charset=UTF-8 to
the Content-Type header for all files of type text/plain or text/html
(IIRC), unless you explicitly specified a different charset for the
requested file.

Without that option (or with it explicitly turned off), Apache will only
append a charset to the Content-Type header, if it was explicitly
specified (eg. using AddCharset).

(Of course it's also possible that someone explicitly defined charset
UTF-8 for .html files.)


Either way, there are a few ways to fix this:
1) If it's caused by AddDefaultCharset, remove the setting. It's
primarily useful if all your files share the same charset (obviously
not the case), or if it is too cumbersome to specify the correct
encoding for all files and you rather want to just override those that
don't have the default charset.

2) Assuming that you have override permissions for FileInfo, you can
override the default, or explicitly specified, charset setting in a
.htaccess file. The following settings should work:

AddDefaultCharset off
RemoveCharset .html
AddCharset ISO-8859-1 .html

You could probably even get along with just the AddCharset directive,
but the above is the safe bet. Add file extensions as required, eg.:

RemoveCharset .htm .html .txt .foo
AddCharset ISO-8859-1 .htm .html .txt .foo

3) If you don't have FileInfo override permission, or don't want to
create a .htaccess file, but have access to the server configuration
itself, you can create a Directory or Location section in the server
configuration containing the same setting as above.

Björn