Re: draft-ietf-tsvwg-iana-ports-09: How we have resolved WG last call comments

[Joe Touch <[e-mail]>]

Hi, Eliot,

On 12/3/2010 4:52 AM, Eliot Lear wrote:
> George,
>
> This is a fair question:
>
> On 12/2/10 6:24 PM, George Neville-Neil wrote:
>> I tried to ask this before, but I think that mail got rejected.  What problem
>> are we trying to solve?
>
> In my own mind, what I would like to see is a modernized set of
> procedures that articulate the principles upon which decisions by IANA
> and their technical reviewers (of which I am one) should be made, so
> that there is transparency.  This helps the applicant in predicting a
> reasonable response prior to even an application.

Well, I'll push back on that. There should be a best practices that 
helps users understand how to do a good job, but I don't think the 
existing expert review process needs detailed codifying. It's intended 
to be flexible. To the extent that there are policies, the general ones are:
	- follow the current specs
	- conserve ports where possible

That's it, in a nutshell.

Regarding whether we should do anything in this doc:

>     * As a matter of practical application, the document does not
>       provide sufficient guidance as to when a System port SHOULD be
>       requested, in contradiction with the goal of transparency.

RFC 1340, which was the first to explain the difference, can be cited or 
quoted:

---

The Well Known Ports are controlled and	assigned by the	IANA and on most
systems	can only be used by system (or root) processes or by programs
executed by privileged users.

---

>     * As a matter of history, one cannot reasonably support an argument
>       FOR OR AGAINST an application based on the existing database.  For
>       instance, why is whois++ (63/tcp/udp) or ups (401/tcp/udp) in the
>       system port range, whereas Radius and Diameter are up in the User
>       range?
>
> The reason THIS document needs to deal with the matter is that it is
> THIS document that is reaffirming the distinction.

I don't think it is; it's just not changing it.

I agree this issue needs debate, but it's independent and can/should be 
dealt with as a separate issue.

FWIW - I just submitted a doc on ports use with some issues that have 
come up in the past, as well as some recent discussions 
(draft-touch-tsvwg-port-use-00.txt). It will be posted to the list shortly.

Perhaps that can serve as a place we can make recommendations *to users* 
on how to use/request ports. Right now it has a lot of detailed history, 
but only mostly placeholders for the rest. Suggestions welcome...

Joe