Re: [Model] 4.5.6 <alias> Element

"Joel M. Halpern" <[email protected]> Tue, 23 Jan 2007 17:59:22 -0500
Newsgroups gmane.ietf.forces
Message-ID <[email protected]>
Set and Get via Alias are supported so as to allow one to set / get 
the entire structure.
Also, while the LFB structures defines the primary location of the 
information, it is entirely up to the CE (or the programmer of the 
CE) what path is used to set / get information.  I can easily imagine 
implementation techniques where it simplifies life to access the 
values through the alias.  (The simplest example is an LFB instance 
browser with set/get capability.)

Mostly, I didn't think it was the model's job to control how the 
values were manipulated.  And there is no conflict, given that all 
the LFBs of an FE are under the control of a single CE.  Yes, I can 
imagine questions arising if we support split control.  But so many 
other issues arise that I see no point in planning for that.

Yours,
Joel

At 11:05 AM 1/23/2007, Jamal Hadi Salim wrote:

>- I think this section would benefit from an example right at this
>point. It also makes it consistent since the other 4.5.x sections mostly
>have  examples.
>
>- The text may be a little confusing.
>Lets say i have an LFBx which uses an alias ifindex. Say the real
>location (target) is in some portLFB:instanceI:a.b.c.d and the value at
>that location is 0x12. Then:
>On LFBx (given i have write permission)
>i) I can change the target to be new path say portLFB:instanceJ:a.b.c.e
>ii)I can SET the value of a.b.c.e (if i have write permissions)
>
>It almost seems to me you should never have the need for #ii (we have
>been using aliases and have had no need to so far) and that you should
>occasionally need to to do #i (for reasons i wont go into we  do it more
>than occasionally).
>
>The text talks about #ii but not about #i.
>
>cheers,
>jamal