note on testing

Thu Feb 15 02:00:00 GMT 2007

Last note on this for today, I promise.
Tom> The problem seems to be an infinite recursion computing a stack
Tom> trace. I'm not sure why.
What happens is that we try to print a stack trace while computing the
inherited thread locals for a new thread. This in turn runs addr2line
for the first time, which starts the new ProcessManager thread, which
computes the inherited thread locals ...
I'm not sure what I want to do about this yet. Maybe run the primary
user thread as a child thread of the initial thread? Then we could
arrange for the ProcessManager internal thread to inherit from the
initial thread. (I think we have a couple other internal threads we
create. ISTR one in Gtk, and one in prefs. Any others? This sort of
thing should probably go into Classpath.)
Or we could start the ProcessManager thread early.
Tom