[NNTP] CAPABILITIES problem!

[Russ Allbery]

Clive D W Feather <clive at demon.net> writes:

> We can change or remove either or both limits. For example:

>   A "keyword" MUST consist only of US-ASCII letters, digits, and the
>   characters dot (".") and dash ("-"), and MUST begin with a letter.
>   Keywords MUST be at least three characters and MUST NOT exceed 16
>   characters.

>     keyword = ALPHA 2*15(ALPHA / DIGIT / "." / "-")

> or:

>   A "keyword" MUST consist only of US-ASCII letters, digits, and the
>   characters dot (".") and dash ("-"), and MUST begin with a letter.
>   Keywords MUST be at least two characters.

>     keyword = ALPHA 1*(ALPHA / DIGIT / "." / "-")

> No other wording changes would be needed. I don't know what the
> implications are of changing the minimum or maximum command
> length. There is no such limit in RFC 977 and I don't know where it came
> from.

I have no idea why there's a limit either.  We've had a limit, as Ken
noted, for the entirety of this process, but I don't know why it was added
in the first place since RFC 977 doesn't have one.

Given that, it seems cleanest to just remove the limit; that's a fairly
minimal change to the document.  It's a more aggressive change to the
standard, however.  Can anyone on the list think of any reason why
removing this length limit entirely would cause any problems with existing
software?

Note that, of course, removing the limit doesn't cause incredibly long
keywords to suddenly appear; someone would still have to define a command
that uses an extra-long keyword.  So the only place that I could see this
even potentially being a problem would be at the generic parsing layer,
and I really don't see how a length limit on keywords helps there in any
practical sense.

(I guess I do see *one* reason for a limit -- the length of commands that
take message IDs as arguments directly affects the maximum message ID
length that NNTP can deal with, given the overall limit on the length of
commands.  But I'm inclined to let people defining new commands that take
message ID arguments worry about that, rather than limiting everything
just for the sake of that one issue.)

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>