I was thinking about this at the weekend. Is it a *MUST* that a server "understands" that %2c is a ,?
AFAIK those codes are part of the url spec. Any server/client/program that handles http must know about % codes. And given that ascii was the basis for character<->hex mappings back then, one can reasonably assume the server must understand that %2c == ",".
http://www.w3.org/Addressing/URL/4_Recommentations.html#z1
The last section deals with prohibited characters. All prohibited characters 'shall' be encoded in the % form. Other characters may be if one so chooses.
http://www.w3.org/Addressing/URL/5_BNF.html#z73
Defines what is/isn't allowed in a URL. -Drew -- gentoo-user@gentoo.org mailing list