----- Original Message -----
From: "Michal TOMA" <michaltoma@...>

: It seems srrange I agree, but the fact is that in the page :
:
http://www.notzai.info/modules.php?name=Downloads&d_op=viewdownload&cid=2
: those URLs are evrywhere...

I just looked at RFC 1738 and it only allows for % encoding.

: Could you direct me to the portion of code I need to check to
see whats
: going on?

Since this is specific to URLs, best place to do checking and
translation would be when a URL is being parsed. Look at
url_parse() in url.c. To make your fix generic enough, before
url_parse does anything, call a function like
url_translate_faulty_html_encoding which will change the urlstr
if there's an "&" followed by ";". Then submit as a patch
through pavuk's sourceforge system...

: What is strange is that it works in one case and not in
another one? The
: first "&amp;" is always translated right...

Yes, that's strange, but could just be a problem with logging.
If this translation happens before url_parse is called then you
can forget what I said above.

Paul