Observation using trn-test76 and INN 2.4.0

[Russ Allbery]

Kiernan, Alex <alexk at demon.net> writes:

> nnrpd is a mess,

Not as bad as innd, but yes.  (ARTpost is a deeply frightening fuction.)
I have hope that we'll get there slowly....

> I'd *much* prefer changing the API to have handles returned from the
> overview API per group - threading then becomes an option.

Amen to that.

At some point, I'd really like to revisit the whole overview and storage
APIs in a fairly comprehensive fashion.  For example, rather than storing
a bunch of information in static variables, OVopen and SMopen should both
return an opaque struct that callers should then pass back into all
overview and storage functions.  The new tradindexed method is already
ready for that when it happens.

I'd also like to change OVsearch to take a struct as an argument and fill
it in with the data wanted rather than taking a whole bunch of arguments
like it does now, basically the struct article as seen in tdx-private.h.

Dunno when we'll get a chance to do all this work.  Maybe INN 2.5, if
we're lucky.  :)  I definitely plan on trying to significantly reduce the
number of configuration parsers we have in INN 2.5.

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

    Please send questions to the list rather than mailing me directly.
     <http://www.eyrie.org/~eagle/faqs/questions.html> explains why.