Re: [netmod] WGLC on draft-ietf-netmod-rfc6991-bis-11

[Jürgen Schönwälder <[e-mail]>]

On Tue, Feb 15, 2022 at 12:12:04PM +0000, maqiufang (A) wrote:

> I have only one comment: It seems that Table 2 doesn’t list all the
> types defined in “ietf-inet-types” YANG module, e.g.,
> protocol-number, ip-address-link-local, ip-address-and-prefix…
> Should this be fixed?

Yes, this should be fixed. When the initial version was produced,
there was quite some concern about consistency with SMIv2 definitions
and this lead to the tables. Meanwhile, I assume the purpose of the
tables is more to provide a quick overview. Hence I propose to split
the tables into (i) tables that provide an overview of all types
defined in the YANG modules and (ii) tables that detail equivalent
SMIv2 types where they exist (short tables). For the overview tables,
I suggest to add some more information. I am thinking of overview
tables like these (keep scrolling, there is more text below):

* ietf-yang-types

  | Typedef               | Type              | Introduced |
  |-----------------------+-------------------+------------|
  | counter32             | uint32            | RFC 6021   |
  | zero-based-counter32  | uint32            | RFC 6021   |
  | counter64             | uint64            | RFC 6021   |
  | zero-based-counter64  | uint64            | RFC 6021   |
  | gauge32               | uint32            | RFC 6021   |
  | gauge64               | uint64            | RFC 6021   |
  |-----------------------+-------------------+------------|
  | object-identifier     | string            | RFC 6021   |
  | object-identifier-128 | object-identifier | RFC 6021   |
  |-----------------------+-------------------+------------|
  | date-and-time         | string            | RFC 6021   |
  | date                  | string            | RFC XXXX   |
  | time                  | string            | RFC XXXX   |
  |-----------------------+-------------------+------------|
  | hours32               | int32             | RFC XXX    |
  | minutes32             | int32             | RFC XXX    |
  | seconds32             | int32             | RFC XXX    |
  | centiseconds32        | int32             | RFC XXX    |
  | milliseconds32        | int32             | RFC XXX    |
  | microseconds32        | int32             | RFC XXX    |
  | microseconds64        | int64             | RFC XXX    |
  | nanoseconds32         | int32             | RFC XXX    |
  | nanoseconds64         | int64             | RFC XXX    |
  | timeticks             | int32             | RFC 6020   |
  | timestamp             | timeticks         | RFC 6020   |
  |-----------------------+-------------------+------------|
  | phys-address          | string            | RFC 6020   |
  | mac-address           | string            | RFC 6020   |
  |-----------------------+-------------------+------------|
  | xpath1.0              | string            | RFC 6020   |
  | hex-string            | string            | RFC 6991   |
  | uuid                  | string            | RFC 6991   |
  | dotted-quad           | string            | RFC 6991   |
  | yang-identifier       | string            | RFC 6991   |
  | revision-identifier   | date              | RFC XXXX   |
  |-----------------------+-------------------+------------|
  | percent-i32           | int32             | RFC XXXX   |
  | percent-u32           | uint32            | RFC XXXX   |
  | percent               | uint8             | RFC XXXX   |
  |-----------------------+-------------------+------------|

* ietf-inet-types

  | Typedef                 | Type         | Introduced |
  |-------------------------+--------------+------------|
  | ip-version              | enum         | RFC 6021   |
  | dscp                    | uint8        | RFC 6021   |
  | ipv6-flow-label         | uint32       | RFC 6021   |
  | port-number             | uint16       | RFC 6021   |
  | protocol-number         | uint8        | RFC XXXX   |
  | as-number               | uint32       | RFC 6021   |
  |-------------------------+--------------+------------|
  | ip-address              | union        | RFC 6021   |
  | ipv4-address            | string       | RFC 6021   |
  | ipv6-address            | string       | RFC 6021   |
  | ip-address-no-zone      | union        | RFC 6991   |
  | ipv4-address-no-zone    | ipv4-address | RFC 6991   |
  | ipv6-address-no-zone    | ipv6-address | RFC 6991   |
  | ip-address-link-local   | union        | RFC XXXX   |
  | ipv4-address-link-local | ipv4-address | RFC XXXX   |
  | ipv6-address-link-local | ipv6-address | RFC XXXX   |
  | ip-prefix               | union        | RFC 6021   |
  | ipv4-prefix             | string       | RFC 6021   |
  | ipv6-prefix             | string       | RFC 6021   |
  | ip-address-and-prefix   | union        | RFC XXXX   |
  | ipv4-address-and-prefix | string       | RFC XXXX   |
  | ipv6-address-and-prefix | string       | RFC XXXX   |
  |-------------------------+--------------+------------|
  | domain-name             | string       | RFC 6021   |
  | host-name               | domain-name  | RFC XXXX   |
  | host                    | union        | RFC 6021   |
  |-------------------------+--------------+------------|
  | uri                     | string       | RFC 6021   |
  | email-address           | string       | RFC XXXX   |
  |-------------------------+--------------+------------|

In future versions we may add a column indicating the status, but
right now all definitions are current, so I rather not add noise.
(Instead of the column 'Introduced' one could have a column 'Revision'
listing the revision date but somehow pointing to the RFC feels more
valuable for modules that we publish in RFCs. And even better would be
to produce such tables from annotations, perhaps the versioning people
solve that problem.)

While putting these tables together, I noticed that we are not
consistent with the naming. We have percent-i32 and percent-u32 and I
think this is pretty neat since the name indicates that these are
signed and unsigned (int32 and uint32) types. We also have hours32,
minutes32, seconds32, etc. but here the name provides no clue whether
the number is signed. Hence, I suggest to use the naming scheme that
is used for the percent types:

  hours32               -> hours-i32
  minutes32             -> minutes-i32
  seconds32             -> seconds-i32
  centiseconds32        -> centiseconds-i32
  milliseconds32        -> milliseconds-i32
  microseconds32        -> microseconds-i32
  microseconds64        -> microseconds-i64
  nanoseconds32         -> nanoseconds-i32
  nanoseconds64         -> nanoseconds-i64

/js

-- 
Jürgen Schönwälder              Jacobs University Bremen gGmbH
Phone: [phone]         Campus Ring 1 | 28759 Bremen | Germany
Fax:   [phone]         <https://www.jacobs-university.de/>