[NNTP] Article number wording

[Matthias Andree]

Russ Allbery <rra at stanford.edu> writes:

> Matthias Andree <matthias.andree at gmx.de> writes:
>
>> I strongly object, and I expect that the better part of the
>> implementations won't follow this.
>
>> I understand that putting up arbitrary limits is tricky as the limit
>> might be chosen too low, but article numbers are THE primary index -
>> introducing arbitrary length integers here is going to hurt big time -
>> implementors and execution speed.
>
> What do you think of an explicit 64-bit limit?

Fine with me.

I can't imagine how we could possibly bump into a 2^64 > 10^19 article
limit - assuming a group with a post per second, and a year to have
10^7.5 seconds this will suffice for 10^11.5 years.

Current projections how long life on planet Earth will be possible range
between 10^9 and 10^10 years, before The Sun will turn into a Red Giant
and swallow the Inner Planets.

Even if we logarithmize these figures because we hopelessly
underestimate the growth, we'll have almost a dozen years to standardize
and implement NNTP 3.

My NNTP software leafnode uses "unsigned long", so if you compile it on
a 64-bit machine (say, UltraSPARC), so I'm not concerned about such a
limit. By the time 64-bit article number will get interesting, 64-bit
CPUs will have become a commodity.

Note #1: I am assuming that humans are going to read NNTP-transported
articles.

Note #2: I don't care the least about binary file distribution or other
forms of NNTP abuse. If the protocol has a known limitation, and this
one is documented, the users will have to make sure they're not bumping
into it through using the protocol unwisely.

-- 
Matthias Andree