RS> The problem is that you barely even test changes, just release and pray.
RS> It aint exactly a terrific idea to just use stuff released like that.

PE> It gets live-tested by the first person to use it.

Soorree, that aint anything remotely like rigorous testing.

PE> I am recompiling Tobruk now, with that bit of code commented
PE> out, to allow what appears to be "routed echomail".  Then I
PE> will recompile it back to normal.  And then THAT IS IT.  NO MORE.

RS> That doesnt sound like a terribly practical approach, because its quite
RS> feasible for other than PKTs generated by PQWK to have that problem.
RS> In other words, even if the latest PQWK is used, you havent allowed for
RS> some other point which doesnt even use PQWK at all doing that blemish.

PE> No, it's only a problem with people who send me packets directly.

And they dont all happen to use PQWK do they ? And presumably
you dont expect Tobruk to only ever get used by you either.

PE> The problem is that the address of the sender is not
PE> 711/934 but whatever crap was in an inserted origin line.

Yes, no argument that there is a problem with embedded origin lines.
How best to DEAL with them tho is another matter entirely tho.

You can for example make a case for nuking the * on one of those
embedded origin lines IN TOBRUK, when its tossing a packet from
people who send you packets directly. And THEN it catches those warts
even when something else is used to prepare the packets than PWQK too.

THATS what robust software is all about. Recognising that in the real
world not all packets can be guaranteed to be absolutely pristine.

In fact for a time Tobruk was producing some which werent
itself. It would have made no sense for system up stream of
you to be silently binning entire PKTs when a wart was seen.
@EOT:

---