nnrpd development

[Russ Allbery]

Jeffrey M Vinocur <jeff at litech.org> writes:
> On Wed, 6 Feb 2002, Matus \fantomas\ Uhlar wrote:

>> Well, I'd like to see parameters -H -T -X in nnrpd :) maybe also
>> something like -i in innd (the parameter should be renamed probably
>> then)

> Hmm.  On the one hand, whatever is actually listening on the socket (INN
> or inetd, for most people) could do most of this stuff.  On the other,
> people still do use nnrpd -D.  Anyone know how hard it would be to pull
> the code for the above out for common use?

Not trivial, since that code in INN is fairly entwined with INN's general
connection handling stuff, which is a world apart from how nnrpd handles
it.  But not incredibly hard, either.

>> Also, a headers config file (which headers to add, which to remove,
>> which to overwrite) would be nice. Or should I do it with perl?

> This could tie in nicely with the header handling reorganization I was
> going to do anyway.  Russ...?

I'm not sure I really understand the scope of the question.  nnrpd should
not, under normal circumstances, be removing any headers except a small
handful that are historically very safe to remove, and the configuration
of what headers to add really should be simplified (we shouldn't be adding
three separate trace headers, for example).  I'm not sure it's a good idea
to make this area more complex....

An injection agent's obligations under at least USEFOR (and I agree with
USEFOR on this point) are to meddle as little as possible.  That doesn't
mean that we should make it impossible to meddle, but it does mean that
I'm fairly comfortable asking people who want to do nonstandard things to
use Perl filters.

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