--On mercredi 23 janvier 2002 17:12 +0000 Leon Brocard <[EMAIL PROTECTED]> 
wrote:

> Vladimir Pastukhov sent the following bits through the ether:
>
>> Andy, I think it's not a good idea to encode high character codes
>> by default since there are languages that do not use latin-1 entities.
>> There should be a new separate filter or a global configuration option
>> or something, otherwise many people would have to change existing
>> templates to achieve original behavior.
>
> However, this does need to be done for Latin-1 entities to prevent
> cross site scripting exploits (something HTML::Entities does but
> Apache::Utils does not, according to a recent thread on the mod_perl
> list). Hmmm.

Is this true if the correct charset is output, e.g.
Content-Type: text/html; charset=iso-8859-1
In this case I believe accented characters do not need to be converted
into entities. If they are then it would significantly increase the
number of bytes being sent to the clients.
If I understand 
http://httpd.apache.org/info/css-security/apache_specific.html
correctly specifying an explicit charset (or using AddDefaultCharset)
defeats cross-site scripting issues.

--
Eric Cholet



Reply via email to