The MPLS WG Archive

[Pankaj K Jha <pkj@cypress.com>]

Eric Gray wrote:

> I disagree on both points for what, IMHO, are obvious
> reasons.
>
> > The date is irrelevant if there is a problem with a specification.
>
> No, the date is not irrelevant.
>
> Progress is accomplished by achieving results, however imperfect
> they might be, and moving on to the next problem.  Progress is
> not attained as a result of an endless search for the perfect goal.
>
> If you see a problem with a specification, I suggest that you write
> a separate specification to address it.  This is under the heading
> of "moving on to the next problem".

> IF, in attempting to write a
> new specification, you uncover reasons why the existing one will
> not be useful to anyone, THEN it is likely that the existing one
> will be discarded in favor of a new, improved version.

>
>

Eric:

I clearly sense a document-generating attitude in "Progress is
accomplished by achieving results, however imperfect they might be",
"moving on to the next problem" and "however imperfect they might be",
and "the existing one will be discarded in favor of a new, improved
version". Unless MPLS has a short life cycle, I think it is still early
enough to be able to initiate a change.

By the way, who is urging us to "move on to the next problem" after
giving a half-baked solution to the current problem? It is always the
person wanting to move on to the next problem. No one else is asking him
to do that. Industry never says - I depend on you to keep generating
document, don't worry how bad they are - for the coming generation will
inherit them and fix them if they want to do it - you just keep typing
away :-)

Your arguments aren't technical at all. And by the way, I do not think
ANY implementer will get confused when he finds IP-specific
implementation details in a separate draft. Neither will his software
implementation change due to one draft now becoming two drafts. So where
is the adverse impact? Let's fix it now, or it will never get fixed
properly.

If we are that determined upon making MPLS to be IP-specific only (and
telling other protocols to go figure how they are going to fit in the
paradigm) then let us just call it IPLS, or some such thing. No use
calling it MPLS.

I do not think removing IP-specific stuff will take such a long time.
I'm sure we'll get plenty of volunteers to do such a job, and it should
take only a few days.

Regards,
Pankaj