El Wednesday 26 March 2008 14:18:58 Paul Kyzivat escribió:
> AFAIK this is not well defined. I believe it must mean one that has
> invalid syntax. But of course the syntax can't be so messed up that the
> remainder of the message can't be parsed.
>
> Loosely I guess I would say that if it starts to match a known header
> but fails in some way, yet can be parsed as an extension header, then
> you should proceed on that basis, treating it as an unknown header.
>
> You could go beyond that, but the ABNF gives no guidance on how. This is
> much like the recovery a parser for a programming language goes through
> so that it may continue to provide diagnostics beyond an error rather
> than simply giving up at the first error.
>
> One thing that is tricky, and which I don't think is done much, is to
> generate a useful response to a malformed request. This can require you
> to generate a response that is itself malformed.


Thanks a lot. In the parser I'm doing I allow known but incorrect headers if 
they are not neccesary (wrong value syntax in Authoriation...).

But if I receive a malformed header field name or hte abscense of ":" I drop 
the entire message. I assume that is not very common to receive "headers" 
like:

  custom header : XXXXXX   <-- invalid because space in the header name.
  CustomHeader asdasdasd  <-- invalid because no ":"

Hope it's enough.

Thanks a lot.
  

-- 
Iñaki Baz Castillo
[EMAIL PROTECTED]

_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors

Reply via email to