Comments on draft-ietf-msgtrk-protocol-00.txt

>1.  Introduction
...
>Using the model document's terminology, it uses active enabling and active requests with request
>referrals.

I couldn't find any referrals in the document.

>4.2.1.  Tracking ID
...
>The Tracking ID is REQUIRED to be the same value as
>used for the Message ID, without the angle brackets, and augmented by a
>colon ":" and a generational counter.  The generational counter will be
>an unsigned integer value that SHOULD start at the value of 0.  For
>example, if the value of the Message-ID: header is
>"<[email protected]>", then the value of the tracking ID
>will be "[email protected]:0".  If a message is ever resent
>or retransmitted for any other reason by an end-user client, then the
>generational counter MUST be incremented by one.  (The generational
>counter MAY additionally be incremented by a small (<10) random number.)

I disagree with proposed solution of making Message-Ids unique.
Requirement to increment by one will lead to duplicates.
And I don't see why random number should be smaller than 10.
I prefer having some random generated sequence of alpha-numeric
characters (maybe with fixed length) instead. At minimum, I think "MAY
increment" should be replaced with "MUST increment" and 10 is replaced
with something bigger.

>11.  Appendix A	-- Sample Code
...
>
>	 MD5Init(&context);
>	 MD5Update(&context, secret, secret_len);
>	 MD5Update(&context, transaction_id, transaction_id_len);

The last two lines should be in the opposite order, because A = H(T + S)

>	 MD5Final(outbuf, &context);

Alexey