Simon Josefsson wrote: > Those protocols should be extended to send this information, since it > is not possible to handle text encoded in multiple encodings > otherwise. That would be fine in principle but mostly useless in practice. The situation with HTTP shows that very clearly. > I don't think it is for UTF-8 to solve these problems, > which BOM partially tries to do. The BOM is one instance of charset metainformation located *inside* the entity (others are the HTML <meta> element and the XML encoding declaration) instead of outside and therefore detachable. Experience has shown such metainformation to be useful and often the only one available. > (HTTP supports this, the current FTP > extensions addresses it partially, lying implementations cannot be > standardized away though.) Indeed, protocol standards have limited powers. In practice, even correct implementations of correctly internationalized protocols cannot convey information that is simply not available to them. Better have belts *and* suspenders in those cases. Regards, -- FrançoisReceived on Saturday, 5 October 2002 12:45:49 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 5 June 2006 15:10:54 GMT