At 16 Apr 2008 12:52:00 +0300, Jari Arkko <[EMAIL PROTECTED]> wrote:
> Alfred, thanks for your review. The stuff that you sent me seems to be
> things that we need to fix. Please prepare a report and send it to
> Heejin. Heejin, please wait for Alfred's fixes before you submit the -14.
>
> Jari
Here we go.
Note 1:
As usual, change bars ('|') in column 1 and ^^^/vvv style pointer
lines are used to precisely point to dangling text or changes
proposed; they are annotation and in no way intended to become
part of the draft. Proposed text is re-`adjust`ed to RFC style.
Note #2:
Numbering of items preserved from discussion with Jari,
thus item (c) now intentionally left off below.
Note #3:
All paragraphs affected by the retrofit requested by
Jari (no MIP4, only dual-stack support) and already
modified by Heejin in the meantime are excluded from
the list below. The same holds for the other small
corrections already requested in this thread.
(a) Section 4.1, 4th paragraph
OLD:
When the mobile node sets the Id-type to 1 in the request, it MUST
include a sub-option with Sub-opt-code 1 which carries the FQDN of
the target network such as "example.com". Note that a single Home
| Network Information option can carry at most one sub-option with Id-
| type 1 in the request.
^^^^ ^^^
NEW.1:
When the mobile node sets the Id-type to 1 in the request, it MUST
include a sub-option with Sub-opt-code 1 which carries the FQDN of
the target network such as "example.com". Note that a single Home
| Network Information option can carry at most one sub-option with Sub-
| opt-code 1 in the request.
^^^^^^^^ ^^^^
or:
NEW.2:
When the mobile node sets the Id-type to 1 in the request, it MUST
include a sub-option with Sub-opt-code 1 which carries the FQDN of
the target network such as "example.com". Note that a single Home
| Network Information option can carry at most one sub-option with Sub-
| opt-code 1.
^^^^^^^^^^^ ^^^^
RATIONALE:
Sub-options do not have an Id-type of their own.
The 6th paragraph of 4.1 and Section 4.3 clearly admit multiple
Home Network Information options in the request with Id-type 1,
but different Home Network Identifiers (target MSPs) in their
(single) sub-option of Sub-opt-type 1. Therefore,
"Sub-opt-code" should be substituted there for "Id-Type".
But the modified last sentence -- per NEW.1 -- equally holds for
the response, if I truly understand the intent of the draft.
Thus, it apparently makes sense to also suppress the trailing
"in the request", as done in NEW.2.
Related PRO/CON considerations:
- The mobile node cannot enforce this condition in the response.
- The 8th paragraph already requests that the MN ignore a HNINF
option with Id-type 1 in the reply which does not contain a
home network identifier (Sub-opt-type 1) sub-option.
I recommend NEW.2, but also have no objections to using NEW.1,
if you have arguments in preference of that.
(Jari also has indicated leaving the choice to the authors.)
(b) Section 4.1, 5th paragraph
OLD:
The mobile node MUST NOT include a Home Network Information option
whose Id-type is other than 0, 1, and 2 as defined as Section 3.1. A
| value of 1 is only available Sub-opt-code in the request.
^
SYMPTOM:
The last sentence is ambiguous.
Does "A value of 1" refer to the Id-type or to the Sub-opt-code?
Hence, does it mean:
[...]
| A value of 1 is the only Sub-opt-available code in the request.
^^^^^
or:
[...]
| In the Request, only a Home Network Information option whose Id-type
| is 1 can carry a sub-option, and this sub-option must have Sub-opt-
| code 1.
DISCUSSION:
In any case, I suggest to separate the clarified sentence from the
preceding one by a paragraph break (blank line), to better decouple
it from the preceding statement that is only related to the Id-type.
Thus, the two alternatives above result in the two proposals (below).
I'd prefer (NEW.1) over (NEW.2) because the latter restates
requirements already expressed in preceding parts of the text
and thus would add redunduncy to the text.
(Jari also has indicated leaving the choice to the authors.)
NEW.1:
The mobile node MUST NOT include a Home Network Information option
| whose Id-type is other than 0, 1, and 2 as defined as Section 3.1.
|
| A value of 1 is the only Sub-opt-code available in the request.
or:
NEW.2:
The mobile node MUST NOT include a Home Network Information option
| whose Id-type is other than 0, 1, and 2 as defined as Section 3.1.
|
| In the Request, only a Home Network Information option whose Id-type
| is 1 can carry a sub-option, and this sub-option must have Sub-opt-
| code 1.
(d) Section 4.1, 2nd-to-last paragraph
OLD:
As described later in Section 4.3, servers attempt to place multiple
options and in the order of preference. When provided with more than
| one Home Network Information options having the different id-types or
^^^^
with multiple sub-options for the same id-type, the mobile node
SHOULD choose the first one that it can employ.
NEW:
As described later in Section 4.3, servers attempt to place multiple
| options and in the order of preference. When provided with multiple
| Home Network Information options having different id-types or with
| multiple sub-options of the same id-type, the mobile node SHOULD
choose the first one that it can employ.
RATIONALE:
Delete the tagged "the" for clarity.
Also tried to improve the readability and avoid a dangling singular/
plural conflict by again replacing "more than one" with "multiple".
(e) Section 4.3, 1st paragraph, last line
Typo; please change "a Id-type" --> "an Id-type"
(f) 4.3, 3rd paragraph, 1st sentence
OLD:
vvvvvvvv
| When a DHCP server receives an Information-request message or it in
the Relay-forward message, it SHOULD check whether the Information-
request includes the Home Network Information option. [...]
NEW.1:
| When a DHCP server receives an Information-request message or it is
vvvvvvvvvvvvv v ^^^
| encapsulated in a Relay-forward message, it SHOULD check whether the
Information-request includes the Home Network Information option.
[...]
or, with more pleasant wording:
NEW.2:
| When a DHCP server receives an Information-request message either
vvvvvvvvvvvvvvvvvvvvvvvv v ^^^^^^
| directly or encapsulated in a Relay-forward message, it SHOULD check
whether the Information-request includes the Home Network Information
option. [...]
RATIONALE:
Sentence is incomplete. I guess that "is encapsulated" is missing
and should be inserted to give proper sense to the sentence. NEW.2
should further improve the readability; I recommend it over NEW.1.
[Heejin: A hint to this garbled/incomplete sentence was intended for
inclusion in my review of -12, but inadvertantly misplaced to point
to another section, and has triggered a change there; I apologize.]
(g) Section 4.3, 5th paragraph:
DISCUSSION:
To clarify the causality in the reasoning, "In case" needs to be
added in front of the second sentence (which introduces a three-
fold alternative discussion).
OLD (abbreviated):
It MUST construct the Home Network Information option(s) according to
| the following logic, and include it or them in the Reply. The
Information-request message includes:
[...]
NEW (with more context, for clarity):
It MUST construct the Home Network Information option(s) according to
| the following logic, and include it or them in the Reply. In case
| the Information-request message includes:
^ ^^^^^^^
o A. Home Network Information option with Id-type 0
[...]
o B. Home Network Information option with Id-type 1
[...]
o C. Home Network Information option with Id-type 2
[...]
(h) Section 4.3, 2nd-to-last paragraph
OLD:
The DHCP server should provide all of the matching home information
in Home Network Information option(s) based on its policy. It may
| return the partially matched results. [...]
^^^^^
NEW:
The DHCP server should provide all of the matching home information
in Home Network Information option(s) based on its policy. It may
| return partially matched results. [...]
^
RATIONALE:
Matching is not considered in detail beforehand, so there is
no point in talking about "*the* partially matched results".
Thus, for clarity and consistency, delete the tagged "the".
(i) Section 4.3, last paragraph
OLD:
There can be several ways that the DHCP server knows the requested
home network information. For instance, as described in [I-D.ietf-
| mip6-radius], a NAS can learn the information via RADIUS during
network access authentication, and the DHCP relay co-located with the
NAS can transfer it to the DHCP server [...]
NEW:
There can be several ways that the DHCP server knows the requested
home network information. For instance, as described in [I-D.ietf-
| mip6-radius] and [I-D.ietf-dime-mip6-integrated], a NAS can learn the
| information via RADIUS or Diameter, respectively, during network
access authentication, and the DHCP relay co-located with the NAS can
transfer it to the DHCP server [...]
RATIONALE:
Other parts of the draft already have been updated to always
mention the DIAMETER work-in-progress in parallel to the
RADIUS work-in-progress as a suitable building block in AAA.
For consistency, this should be done here equally.
Kind regards,
Alfred Hönes.
--
+------------------------+--------------------------------------------+
| TR-Sys Alfred Hoenes | Alfred Hoenes Dipl.-Math., Dipl.-Phys. |
| Gerlinger Strasse 12 | Phone: (+49)7156/9635-0, Fax: -18 |
| D-71254 Ditzingen | E-Mail: [EMAIL PROTECTED] |
+------------------------+--------------------------------------------+
_______________________________________________
Gen-art mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/gen-art