Re: [Teas] WG Last Call: draft-ietf-teas-yang-te-29

[t petch <[e-mail]>]

On 11/04/2022 12:17, t petch wrote:
> On 21/03/2022 13:28, Lou Berger wrote:
>>
>> All,
>>
>> This starts working group last call on
>> https://datatracker.ietf.org/doc/draft-ietf-teas-yang-te/
>>
>> Given the size of the document and this week's meeting, this will be an
>> extended LC. The working group last call ends on April 13th.
>> Please send your comments to the working group mailing list.
>>
>> Positive comments, e.g., "I've reviewed this document
>> and believe it is ready for publication", are welcome!
>> This is useful and important, even from authors.

I have not reviewed the document - it is too complex, largely due to the 
way in is constructed - but have seen enough to think, 'Not ready for 
publication'.

Security has the concept of number used once, soomething which makes it 
harder for anyone else to know what is going on.  The equivalent in YANG 
is 'grouping used once' which I often see YANG Doctors point out and ask 
to be removed since they increase complexity, reduce ease of use. This 
I-D does just that and goes further, nesting groupings extensively which 
makes it even harder, verging on impossible, to understand.

I start at a high level, as in Section 5, intending to 'drill down' to 
the lower levels, leaf and its type and description, to see if they make 
sense, if they do what is wanted, and the highest level is good; the 
English needs some tweaking but nothing I am unclear about.

With s.5.1.1, I get uncertain - where is the description of path 
in-segment and path out-segment that feature in the tree diagram?

With s.5.1.2 I know that something is wrong.  The YANG snippets are not 
those of this I-D; they would help if they did but they relate to a 
different I-D.  Thus the YANG has, buried in a grouping used once,
leaf description
leaf admin-state
leaf operational-state
s.5.1.2 has
description
operational-state
encoding
a related, but different, module.

Likewise, the YANG has bidirectional after dst-tunnel-tp-id in another 
grouping used once, not in this snippet.  At which point, I cry 'Hold, 
enough'; I cannot trust what I am reading.

Another problem, besides difficulty of understanding, with grouping used 
once, is the difficulty of relating the different parts of the I-D, 
description, tree diagram, YANG and so on.  The obvious solution is 
comment lines showing where globals ends and tunnel starts, in the YANG, 
in the tree diagram, in the introductory tex.  That is pointless here 
since in the YANG, they are adjacent 'uses' and at a more detailed 
level, globals, tunnels and so on are all jumbled up.

The IETF has abolished the page number so you will have 40 
undifferentiated pages of tree diagram followed by 40 undifferentiated 
pages of YANG, but with objects in a different order, courtesy of 
grouping and grouping used once.

Tom Petch

> It has been a few years since I reviewed this so the
> Copyright 2019
> seems about right.
>
> However, reading Adrian's comments brought back memories and indeed, my
> notes say 'three e-mail of comments' which suggests to me that they were
> never acted on:-)
>
> So
> - there are ten references in the YANG to RFC that do not appear in the
> I-D references (used to be four)
>
> - abbreviations need expanding, I suggest in s.2 with a separate heading
> to the current s.2
>
> - http: needs to be https:
>
> - the TLP are out of date (that's new)
>
> - the format of the tree diagram is clumsy - identifiers are split in
> the middle, across lines
>
> - revision should be 'Initial ..' not 'latest'
>
> - /constraitnt/constraint/
>
> I will plough on - my target is by Easter, ie COB 14th April
>
> Tom Petch
>
>>
>> Thank you,
>> Lou (Co-Chair & doc Shepherd)
>>
>>
>>
>>
>> _______________________________________________
>> Teas mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/teas
>> .
>>