Posting over innd/lc.c

[Russ Allbery]

Jeffrey M Vinocur <jeff at litech.org> writes:

> Right, that's what we figured.  But I can't for the life of me figure
> out where it is.  (It was changing that MyAddress assignment to using
> memset to zero the struct that caused the problem.  But I don't get it.)

I don't either, particularly since that should really be equivalent.
That's deeply bizarre.

> Agreed.  Except that it we need to track down the existing code to
> expunge, hopefully, and that we're having trouble doing.

I have a sneaking suspicion that it's somewhere deep in the incoming.conf
parser in rc.c, and that's due to be replaced at some point, so I wouldn't
worry too much about this.  I at least can't figure out that code well
enough to be sure that's not where it is, and that's the only place I can
figure out it could be.  Unless something else is going on in ARTpost,
which needs to be broken into at least a dozen separate functions.

> On the plus side, adding a check for ==0 to the top of RCcanpost and
> returning 1 in that case, bypassing incoming.conf, causes everything to
> work.  So it's fixed for the time being, in which case I think it's time
> to commit this and we can work on figuring this out more as we like.

> Is it worth adding to TODO?

Nah, what really needs to happen is that rc.c needs to be basically
rewritten using the new parser, and we'll get it as part of that.

Yeah, go ahead and commit.

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