RE: Comments on draft-adid-urn-00

[Jarrett Wold <[e-mail]>]

Thanks Dale -
I will review and get back to you soon. I'll get this submitted as an
Internet Draft as well.

I appreciate the feedback.
Jarrett

-----Original Message-----
From: Dale R. Worley [mailto:[e-mail]]
Sent: Monday, March 14, 2016 10:44 PM
To: Jarrett Wold <[e-mail]>
Cc: [e-mail]
Subject: Comments on draft-adid-urn-00

Here are a bunch of comments on draft-adid-urn-00.  Overall the concept
seems fine, but the document needs cleaning up.

The title contains "Adverting".  Have you run the draft through a spelling
checker?

      The following is an example of an Ad-ID Identifier in its
      canonical representation:

             ABCD0001000

      The canonical URN representation of the same Ad-ID Identifier is:

         urn:adid:ABCD0001000
         urn:adid:ABCD0001000H


      The 32-bit unique representation of the Ad-ID is:

         urn:adid:cuid:36faeaba

This example is difficult to follow.  The first item is straightforward,
as it is an 11 character string.  But the next item isn't clear:  by
definition, a "canonical" representation is one representation chosen from
among many.  You can say that both

         urn:adid:ABCD0001000
         urn:adid:ABCD0001000H

are URN representations of the Ad-ID identifier, but only one of them can
be canonical.  Also, it would be helpful if you explained why both URNs
correspond to ABCD0001000, or at least provide a reference to some
description which will explain it.

But, reading http://www.ad-id.org/how-it-works/ad-id-structure, it seems
that ABCD0001000H is not the *same* Ad-ID as ABCD0001000, but rather the
Ad-ID for the corresponding High Definition version of the asset.  You
need to clarify this so that the example is not misleading.

It is not clear how the further "32-bit unique representation"
corresponds to the preceding three representations.  It's reasonably clear
what "32-bit" means as the URN contains a sequence of 8 hex digits, but
it's completely unclear what is "unique" about it.

Remember, this is the introduction, which should be readable by people who
don't already understand the technology, including people who don't
already understand Ad-IDs in all their forms.

There's some general trouble about the indentation of text.  For instance,
it seems that "The identifier structure is as follows:"
should be 4 spaces to the left, "Non alphanumeric ("special") characters
and spaces are not valid within an Ad-ID." should have a blank like above
it and be moved 3 spaces to the left, and "A Canonical Full Ad-ID ..."
should be moved 1 space to the left.

This text has some redundancy that should be removed:

          A Canonical Full Ad-ID Identifier shall conform to the syntax
          specified below using ABNF (as specified in IETF RFC 5234):
          The URN representation URN-ADID of an Ad-ID Identifier
          conforms to the syntax (expressed using ABNF (as specified in
          [RFC5234]):

The production

          adid_prefix = (ALPHA / %x30-39) 4*alphanum
               ; first character not zero

is incorrect on two counts:  As written, the first character can be zero
(which ix %x30).  The strings generated are 5 characters.  What you want
is

          adid_prefix = (ALPHA / %x31-39) 3*alphanum
               ; first character not zero

The instances of ALPHA and DIGIT might be intended to be "alpha" and
"digit".  Or do you mean to reference the productions ALPHA and DIGIT in
RFC 5234?  If so, you should eliminate the productions for "alpha"
and "digit" in the draft.

The BNF does not show how the "urn:adid:cuid" URNs are generated.  As the
BNF now stands, those examples are invalid.

      Ad-ID URNs are resolved via URN resolvers run under Ad-ID's
      responsibility.

It would be helpful here to provide a pointer to the documentation for the
resolvers.

      The validity of an URN-ADID can be checked using Ad-ID's web
      services.

Ditto.

Dale