[NNTP] Article number wording
Ade Lovett
ade at lovett.com
Tue Jul 19 01:30:32 PDT 2005
Matthias Andree wrote:
> I understand that putting up arbitrary limits is tricky as the limit
> might be chosen too low
The primary problem with limits is the requirement for the associated
wording of the text around what happens when (not if) those limits are
reached.
> but article numbers are THE primary index
Absolutely, utterly, and totally, *wrong*. Article numbers are most
certainly *an* index, but they are not *the* index. The fact that a
given article can have completely different numbers for each of its
groups across every single Usenet server should make that clear.
*THE* primary index of Usenet is the message ID. There is even language
in the text to indicate that it SHOULD (RFC-speak) be globally unique.
I can go to pretty much any Usenet server on the planet, and request
<abc.123 at example.com> and get the same article. The same cannot be said
for article #123456789 in news.software.nntp.
> introducing arbitrary length integers here is going to hurt big time -
> implementors and execution speed.
I see absolutely no compelling reason to move away from what we have
right now in the final draft that is due to become a *Proposed* Standard
in the near future.
We have spent far too much time already in getting this out and *that*
is hurting implementors far more than 32 vs 64 vs 128 vs n-bit ethereal
article number representation.
By all means continue to discuss this, and other, issues that have
arisen. Remember, we can always issue a new document in the future with
more stringent guidelines (and maybe it won't take 7+ *years* like this
one).
But for now, we have a good solid document. Sure, it's not perfect,
nothing ever is, but it's a damn sight better than what is out there
right now.
Things can always evolve. It is their nature.
For the record, I strongly object to any specific mention of
32/64/whatever-bit text in the document as it stands now. I firmly
believe we would be doing a major disservice to the Internet community
at large by delaying publishing this over one of a myriad of
implementation issues.
-aDe
More information about the ietf-nntp
mailing list