Re: Reminder: WGLC Announcement for draft-ietf-tsvwg-iana-ports-08 - 26th November 2010

[Eliot Lear <[e-mail]>]

Magnus,

There were only two issues with that draft:

1.  I mangled a few issues together (like how to keep port assignments
current, and the value of SRV records);
2.  I ran out of time to do the work at that time.

I would be happy to revive a simpler draft but I think it's better that
you just deal with the matter now, so that we don't have to go through
multiple iterations of this.

As I recall, though, nobody really had a problem with dropping the
distinction.  It's only there in some UNIX flavors; and the only real
issue is on multi-user systems where the port could conceivably be
grabbed by someone.  Realistically, that's not a concern because if it's
important, there is something listening from start-up.

Eliot

On 12/1/10 10:01 AM, Magnus Westerlund wrote:
> WG, Paul and Eliot,
>
> I will respond in detail. However, I do need to do a bit of history
> digging and and go review the fairly big discussion that was held on the
> [e-mail] list back in 2006 that was started by Eliot Lear.
>
> The threads of relevance appears to be:
> https://www.ietf.org/ibin/c5i?mid=6&rid=49&gid=0&k1=933&k2=30752&tid=1291193700
>
> https://www.ietf.org/ibin/c5i?mid=6&rid=49&gid=0&k1=933&k2=31742&tid=1291193820
>
> Elliot's individual draft:
> http://tools.ietf.org/html/draft-lear-iana-no-more-well-known-ports-02
>
> Maybe Eliot can summarize what was the reason his effort failed to gain
> sufficient traction to be published?
>
> Cheers
>
> Magnus
>
>
> Paul Hoffman skrev 2010-11-24 17:27:
>> At 5:17 PM +0100 11/24/10, Magnus Westerlund wrote:
>>> Paul Hoffman skrev 2010-11-22 23:14:
>>>> In general, this document seems fairly worthwhile. I have a two reservations, however:
>>>>
>>>> - There is no justification for retaining the differentiation between System Ports and User Ports. Given the wide disparity in assignment rates, I would have thought that this would be a good time to say "there is no longer a difference". The text in 8.1 doesn't explain the difference in a way I could discern. At a minimum, this needs to be covered in much more detail in sections 7.1 and 7.2.
>>> My personal view is that I agree that there really are no significant
>>> difference between the two ranges. There has traditionally been a
>>> perceived difference between the two ranges.
>> That is only because *we* said there was a difference.
>>
>>> Also, isn't there still
>>> some difference in what rights are needed on a number of unix systems to
>>> install a listener?
>> Not in any sane system, no.
>>
>>> So I think the difference is in peoples heads. The
>>> registration rules do require you to clearly motivate why you should be
>>> given a port in the system range.
>>>
>>> In chapter 6, there is the following text:
>>>
>>>    Such confirmation of intended use is
>>>    especially important when these ports are associated with privileged
>>>    (e.g., system or administrator) processes.
>>>
>>>
>>> For the difference in allocation rates there is a reason why there such
>>> a low rate for "System" ports, and that is due to the high bar that
>>> already is set by the port expert reviewers.
>>>
>>> We are trying to focus on getting the new registry and its structure in
>>> place. Rather than changing all details, like if the system port range
>>> should be removed. There was previous discussion on this in IETF without
>>> any consensus so we haven't been interested in driving this.
>> If not now, while the registry is open, when?
>>
>>> I think removing the system ports range is beyond our intentions with
>>> this document. Secondly, we can try to clarify the difference between
>>> system and registered range.
>> In the IETF, tomorrow's tomorrow is never.
>