sync errors (was Re: xtconq status)

[Jim Kingdon]

> When I run two copies of the tcltk version on the same machine I never
> see these sync errors. That's why they have been so difficult to
> debug.

Well, assuming -seq and no AI, I've seen them at the end of a turn
(and only once every 20 or 40 turns).  Maybe putting sleep(1) in some
places related to the end/start of a turn would help in reproduction?
It does seem to be timing dependent (type too quickly at the wrong
time, or some such).

If the save/restore/checkpoint code worked for networked games, it
would merely be an annoyance.  But last I heard, that didn't work
(restore has no way of assigning players to sides in this case).