Re: [Ntp] [EXT] Re: Re: Experimental/Private EF area: was Re: Registries document

Hi!

My last sentence on "02" being the message length was nonsense; it's the field type as I understand it. Actually I'm not surprised that reading that confused me quite a lot...

Regards,
Ulrich

-----Original Message-----
From: ntp <ntp-bounces@ietf.org> On Behalf Of Windl, Ulrich
Sent: Tuesday, August 1, 2023 8:42 AM
To: Miroslav Lichvar <mlichvar@redhat.com>; Salz, Rich <rsalz@akamai.com>
Cc: Harlan Stenn <stenn@ntp.org>; Harlan Stenn <stenn@nwtime.org>; ntp@ietf.org
Subject: [EXT] Re: [Ntp] Re: Experimental/Private EF area: was Re: Registries document

Hi!

Actually I think the big mistake made in the past was to add the whole word into the IANA registry when the word is interpreted to consist of four or five fields.

IMHO the correct option would have been to register the "Code"s and maybe the "Field Type"s (as per RFC 5906, Figure 7: NTPv4 Extension Field Format).

BTW: The IANA registry lists 16-bit numbers as "Field Type", but tn the RFC the "Field Type" is just 8 bits; so maybe it's "R" + "E" + "Code" + Field Type" actually.

But then I wonder whether "02" is actually the "Field Type" (what's type 2 , BTW?), or the "autokey version 2". Harlan still denies it's a mess...

Doing it correctly the actual table would have been much shorter. Maybe compare it to the TIFF (Tagged Image File Format) specification where tags, types, and lengths are clearly separated.

    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1

   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   |R|E|   Code    |  Field Type   |            Length             |

   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Also for example taking "0x0302

Cookie Message Request" the length should be 2, but I wonder how to put "host public key encoded with ASN.1 syntax" in that size.

(If the last part formats badly: It's Outlook when pasting text from a PDF)

Regards,

Ulrich

-----Original Message-----
From: ntp <ntp-bounces@ietf.org> On Behalf Of Miroslav Lichvar
Sent: Monday, July 31, 2023 4:41 PM
To: Salz, Rich <rsalz@akamai.com>
Cc: Harlan Stenn <stenn@ntp.org>; Harlan Stenn <stenn@nwtime.org>; ntp@ietf.org
Subject: [EXT] Re: [Ntp] Experimental/Private EF area: was Re: Registries document

On Mon, Jul 31, 2023 at 02:12:21PM +0000, Salz, Rich wrote:

> As we clearly do not yet have WG consensus on changing the private-use field types, I am not making the change yet. Also, because the text was wrong.

> 

> The current request is 0xF000 through 0xFFFF inclusive which is 4096 values. Harlan proposed 16 partitions, 0x000E thru 0xFF0e, 0x001E thru 0xFF1E, ... and 0x00Fe thru 0xFFFE. If I got the math right, that's also 4096 values.

> 

> There is no "partitioning" within the private-use extension field values.  It's a flat numbering space that IANA stays away from.

I think the idea was to avoid potential implementations of the current

RFC 5906 looking only at the low 8 bits of the type (the "version"

subfield) to misdetect experimental EFs as Autokey. The suggested 0x???E

values don't intersect with the Autokey values 0x??02. But as we are

fixing the registry and RFC 5906 to use the ntpd values (as the RFC

was supposed to do), 0x???E would intersect with the 0x(0|4|8|C)2??

values. The 0xF??? values which are now in the draft is fine.

-- 

Miroslav Lichvar

_______________________________________________

ntp mailing list

ntp@ietf.org

https://www.ietf.org/mailman/listinfo/ntp