Re: [Netconf] Confirmed commit

[Jonathan Hansford <[e-mail]>]

On 2013-09-19 21:15, Juergen Schoenwaelder wrote:

> On Thu, Sep 19, 2013 at 09:07:17PM +0100, Jonathan Hansford wrote:
>
>> After about two hours digging around in the NETCONF archive and then 
>> a
>> Google search on "confirmed commit", I've come to the conclusion
>> confirmed commit has come out of the JUNOS "commit confirmed" 
>> command.
>> My guess is the JUNOS command is shorthand for "revert this commit
>> after the timeout unless it is confirmed"; not exactly clear. 
>> Indeed, I
>> would have thought "commit confirmed" would be the command to 
>> confirm
>> the previous commit, though obviously that would overload the 
>> meaning
>> of the original "commit" command since it would not persist without 
>> the
>> confirmation. But "confirmed commit" implies the commit has been
>> confirmed, not that it needs to be confirmed. Isn't this confusing 
>> to
>> anyone else?
>
> There is the ':confirmed-commit' capability, there is the 'confirming
> commit' and there is the initial commit which is sometimes called the
> 'confirmed commit'. I assume it is this last one term you believe is
> confusing since it is the 'to be confirmed commit'. Perhaps we could
> have picked a better term - but as long as the description is clear
> (section 8.4 of RFC 6241), I think we are in a good enough state.
>
> /js

Thanks, you have confirmed (no pun intended) my reading of the RFC and 
I will use your definition of it being a 'to be confirmed commit' in my 
document.

Jonathan