None of the machine limits looked likely, and since it seemed to
happen in the same point in the file, I managed to 'manually'
restart it near that point, and after a pause in the output
stream occurred, I hit ^C - where upon I got several pages of
stack trace from python indicating that I have a *slight*
recursion-gone-wild bug in my python code.
Another tribute to the benefits of interactive programming:
not only had I THOUGHT that my code had been pretty well
tested, but it successfully ran on about 200,000 blocks
of a 300,000 block file, before it ran into the "right"
test case.
It WOULD have been nicer if Python could have intercepted
the problem and given me a nicer message, without having
had to force the stack dump with ^C ( i.e. at that point,
I had already sort of guessed what the problem must be,
and was looking for the where. ) but from the replies I
got, it looks like there is nothing Python can do about
catching the Kill signal. [ fodder for one of the comp.unix
news groups: should unix first try signaling with a
"catchable" signal before trying the final irrevokable kill ?
i.e. "STOP WHATEVER YOU"RE DOING", ... "NO - I *REALLY* MEAN IT!"
I mean, that *IS* the typical mode of parent/child interaction!
Isn't it? ;-) ]
- Steve M.