ietf-nntp LISTGROUP

[Russ Allbery]

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

>>  * LISTGROUP is not actually that generally useful enough of a command to
>>    put a lot of effort into reinventing and force client upgrades to use
>>    the new command.  My intuition is that not many people use LISTGROUP,
>>    since one can get the same information from OVER and the client usually
>>    has to send OVER anyway.

> The archives (back from 1996) disagree with you. People wanted LISTGROUP
> because it's much simpler/faster than using OVER or HDR to get the same
> information.

It may be simpler.  I find it dubious for it to be faster.

I'm opposed to the proposal to add a new command with LISTGROUP
functionality.  The fiddling that we've already done with it is still
backwardly compatible.

> Doing them all isn't hard. If I grep for '^ *[1-5][0-9][0-9] ', which
> ought to catch all the response codes in the response sections, then
> tidy up and sort | uniq, only the following duplicates occur:

No, you missed my point.

The response codes for NNTP currently follow very little useful pattern
and don't have well-defined distinctions between such things as temporary
and permanent errors.  That's the sort of fix that would actually make a
difference for client authors.

I agree that it's minorly annoying to have the same return code used in
two different contexts, one multiline and one not, but I don't think that
LISTGROUP is important enough to go to this much effort to fix it.

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