Ugh. I think I see what's wrong -- there was a change in the behavior of
FXTable::setTableSize() in the latest FOX version (1.0.26) that I'm not
properly accounting for. As a result, some of the FXTableItem instances (in
Ruby land) end up holding "dangling" pointers (i.e. pointers to C++ objects
that no longer exist). I've done some initial testing of the fix and now
your bug3b.rb runs to completion. The memory use also stays constant which
suggests that the GC is behaving properly (no leaks or anything).
Let me do some more wrap-up and try to issue a bug fix release this
afternoon. Thanks for your patience...
ยทยทยท
On Fri, 08 Nov 2002 08:50:57 -0800, Joel VanderWerf <vjoel@PATH.Berkeley.EDU> wrote :
I've got bug3.rb, my third attempt to isolate the problem, this time
with 2002-10-30. It's not as short as the earlier bug.rb and bug2.rb,
but OTOH the bug happens sooner. There seems to be an inverse relation
between code size and how long it takes the bug to appear, so I will
stop whittling for now.