PE> so is Christmas a good time to do your own testing on PQWK so that
PE> you can be happy that it is operating properly and then switch to it?

Dunno. What I was planning to do was a complete redo of the whole
operation at my end, mainly to make it a lot safer and completely
fail safe. Thats obviously a rather bigger job and I havent basically
done anything much except think about it.

One of the best ways to make a system like that bullet proof is to
actually process the data twice, once with the current version, once
with the previously used version. Say when processing the REP into
a PKT. AND automate a comparison between the two PKTs. Then with a
new release you can run for a bit with the previous version still
the one which produces the PKT which is used, then when you havent
seen any problems for a bit, swap that around so its the PKT produced
by the new version which is actually exported to you. Still keep
comparing the output of two versions of the code in case some
relatively uncommon thing shows up a wart.

As you can see, its not trivial and I havent done it yet.

I was also thinking about adding MSGIDs, making the netmail more robust,
and some other stuff like that too.

--- PQWK202