Jonathan, Thanks for your response. Please see embedded.
Jonathan Rosenberg wrote:
>
>
>
> > -----Original Message-----
> > From: Shail Bhatnagar [mailto:[EMAIL PROTECTED]]
> > Sent: Tuesday, February 27, 2001 9:04 AM
> > To: Jonathan Rosenberg
> > Cc: '[EMAIL PROTECTED]'; [EMAIL PROTECTED]
> > Subject: Re: [Sip-implementors] the meaning of "copy"
> >
> >
> > > >
> > > > It seems to me that byte-by-byte is more robust and easier to
> > > > compare in
> > > > terms of "From" and "To" headers. What are the potential
> > reasons that
> > > > somebody wants to modify these two headers in addition to the
> > > > "Tag" param?
> > >
> > > The problem has to do with storing whitespace on other
> > formatting > >characters
> > > when parsing. I'd like to be able to parse a header,
> > discard any LWS, and
> > > reconstruct the header. This should be "equivalent" based
> > on concrete
> >
> > I thought URLs cannot have whitespace, so I am not convinced
> > about a proxy/ua messing around with From/To except appending
> > a tag.
>
> There is LWS in the To/From header in several places:
>
> From: me <sip:u@h>;tag=2
> From: me < sip:u@h>; tag=2
> From: me < sip:u@h > ; tag=2
>
> are all equivalent. They would not match under bytewise comparison of
> To/From.
I understand, but I thought we only care about From URL,
possible From tag, To URL and to tag. The last component -
to tag is somewhat tricky. I am not sure how creative
hashing functions will help and am not going to ask that,
but I needed a statement/sentence in the SIP spec about
inclusion/non-inclusion of To header tag when computing
the transaction hash.
>
> > While we are here, I posted some questions several times on
> > the sip-implementors about transaction identification, but
> > nobody responded.
> > If To header tag is included in the hash computation, then
> > ACK for a INVITE will identify a transaction different from
> > the original INVITE. This is acceptable for an ACK for a 200,
> > but not for non-200 ACK ( proxy has to stop response retx timer).
> > If To header tag is not included then 2 different PRACKs will
> > hash to the same transaction.
>
> There are several ways to handle this in an implementation. Its a question
> of being creative with hash table structures.
>
> -Jonathan R.
> ---
> Jonathan D. Rosenberg 72 Eagle Rock Ave.
> Chief Scientist First Floor
> dynamicsoft East Hanover, NJ 07936
> [EMAIL PROTECTED] FAX: (973) 952-5050
> http://www.cs.columbia.edu/~jdrosen PHONE: (973) 952-5000
> http://www.dynamicsoft.com
_______________________________________________
Sip-implementors mailing list
[EMAIL PROTECTED]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
