Re: Control characters in log message cause failure

I wrote:
> Julian Reschke <julian.reschke@gmx.de> writes:
> > Wrong. CDATA doersn't change anything. It's still the same set of
> > characters that are allowed (see for instance
> > <http://greenbytes.de/tech/webdav/rfc3470.html#binary>).
>
> So XML CDATA cannot represent arbitrary strings of bytes, by using
> escaping?
>
> /me sits stunned in his chair

Sorry, I'm not being very clear about things.

Obviously, we can encode anything we want using Base64. So it's not
so surprising that XML's CDATA rules are what you (and that helpful
reference) say they are. It's not limiting what we can do, it's just
making us jump through another hoop to do it.

So my pseudo-code from before should really look like this:

   /* sending a log msg from server to client */

   if (is_valid_utf8 (msg)) {
     if (svn_xml_is_xml_safe (msg)) {
       send_message (escape_xml (msg));
     }
     else {
       send_message (base64_escape (msg));
     }
   }
   else {
       send_message (fuzzy_escape (msg));
   }

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Dec 1 21:17:41 2004