Re: Play testing notes (CVS, 2002-03-10 version)

[Hans Ronne <hronne at pp dot sbbs dot se>]

Thanks a lot for the feedback! To bad things are not working perfectly yet,
but that's what pre-release testing is all about.

>At some point, we would receive the following error.  On my
>display (I was host), in the message section of the main window,
>I would see something along the lines of
>
>        Checksum changed 6835 -> 6832 when numdone == 0
>        Error from #2, "checksum 2 6838 6835"

OK. Any idea what triggered the checksum errors this time? And did you get
errors also if the ai was running either or both sides? I'm asking because
my own error-free game testing, albeit on a single machine, was mostly ai
versus ai. Furthermore, we have already found (and fixed) one synch bug
that was specifically caused by human action: rapid double-clicking on the
same cell.

If you also could test mplayer versus mplayer over a long-distance
connection and see if there are any checksum errors, this would be very
helpful, since it would tell me if the remaining bug(s) are in the human
interface code or in the network code itself.

I also wonder what you mean by "at some point". After 5 turns or 30 turns?
And did you play the game in sequential or non-sequential mode? Was the one
game you almost completed free of errors?

>        Warning:  packet A refers to missing unit #34 [...]
>        Do you want to continue?

I now know what is causing this bug. See the latest ChangeLog. It happens
when you iterate over all units (dead or alive) and tell them to do
something such as force_replan.  Telling a dead unit to do something is in
itself harmless, and does not have any consequences in single player mode.
However, the sanity check in the network code sees that something is wrong.
We could probably just disable the warning, but I will go through the
kernel code and expurgate all instances of this bug just to make sure.

In any case, I don't think this bug was the reason why you got more
checksum errors. It is rather the first error that triggers more as
differences accumulate.

Thanks again for the feedback.

Hans

P.S. Those saved game files might be useful, if you could send them to me.

Hans Ronne

hronne@pp.sbbs.se