More modification to RIPE-81++

> Jessica Yu <jyy at merit.edu> writes:
  >* 
  >* I just came back from vacation so I have not had chance to go through
  >* all the related messages yet.  From what I understand is that we all agreed
  >* to make the <local-rid> field optional and because of changing the ordering
  >*  of 
  >* <local-rid> field which cause some ambiguity, we want to change the 
  >* requirement from 'optional' to 'mandetory'.  This sounds really odd.  Can  
  >* we do something to still keep it 'optional'?
  >*
>Why ??? - See Jimi's message. He had assumed they were always
>mandatory.

I did not see Jimi's message which mentioned that he had assumed them mandatory.
Am I missing his message?  

In fact, Merit's extention syntax which has been used for the RRDB pilot definedthem as optional and as I understand that interas-in/out is pretty much the
mirror of what's defined there.  So I always assumed it optional and we agreed
at Toronto meeting that it is optional.  Now again, I think we act too casually
to make it madatory because we decided the ordering. 

>* Also, this (optional vs mandetory) seems to be a pretty big change.  Since 
  >* two
  >* of the co-authors were (and one is still ) on vacation, would you please wa
  >* it 
  >* for them to come back before making the 81++ final?  Elise will be back on 
  >* 8/31.
  *
>This is not final. Read the top, it STILL says DRAFT. Final decision will
>be made at the RIPE meeting. But I do not see any good argument for it
>being optional. The point is we want it so you SHOULD know both ends.
>As to your comment about waiting for co-authors this is fine in
>principe. However, if we waited for every co-author to not be on
>vacation  before doing anything we would not get anywhere with this. I
>stress again that this process has gone over by several months from
>our (yes - the co-authors) agreed timescales so please bear this in mind.

Hay, you had your vacation, don't you allow V-days for others? :-)  

>Please give convincing agruments as to why you want it optional and
>not do this to try to change the ordering. This has been settled in
>the decided way.

To me, to keep them optional overweights the ordering so trust me that I
won't use it to change the ordering. 

One of the reasons of making it optional is to avoid to have the interas-in/out
include unnecessary information.  It does not reads well, occupies storage
space and is user unfriendly.  Laurent may point this out already: most of the 
cases, one does not need to list both the neighbor rid and local rid to
specify the policy.  Why make them put them down even if that is not needed?  
I really do not see the point.

Let's make some effort to keep them optional.  E.g. to still keep the order
but add tag on it to identify the field something like:

	L:<local-rid> R:<neighbor-rid>

I really do not like this.  I think the format which is not chosen ( 
<local-rid> [from] ASxxx <neighbor-rid>  ) use the position to imply the
local or neighbor which saves tag.

						--Jessica



-------- Logged at  Tue Aug 30 22:44:50 MET DST 1994  ---------