As said, the program is exiting because it gets too many errors. update_button is called whenever the button is pressed/activated. If you are using a button/plate, then all this takes is for some object (of appropriate weight) be be on the space. Now what the error really means is that one of the objects that had the same connected value has 'disappeared'. disappeared could be any of various causes -a monster that was killed, or some other destroyable objects. As currently implemented, connected objects must be permanent. My guess is you have some objects with connected values that are not permanent (note it doesn't make a difference code wise that these destroyed connected objects may not do anything - just anything with a connected value is put on the connected object list at load time). Tom Barnes-Lawrence wrote: > > I've pretty much cornered it now, but I'm not sure what to make > of it- my bad or CF's? > Further experimentation had yet another server crash, but approx > 20 minutes after logging in.So I figured it was almost certainly > to do with map resets. As I'd put in a unique-item tile with > machinery just before the 1st crash, I figured that may be involved. > ... > And then I realised that there *had* been a logfile after all, but > it'd been dumped in my root directory. (Mebbe you should change > compile-time defaults for that?) > Looking through to the appropriate times, I found in each occasion: > resetting map tomble/prototypes/keep_prot-pt6 > Internal error in update button(pedestal) > Internal error in update button(pedestal) > Internal error in update button(button) > Internal error in update button(plate) > Internal error in update button(plate) > <etc etc etc etc> > Fatal, too many errors > (Sorry I'm paraphrasing here, logs are on other machine which is > switched off.If you want an exact quote I'll look tomorrow) > SO you get the point, there's no actual crash-bug as such, it's > just bailing out because of the errors, fair enough.