A Thursday 26 June 2008 03:48:14, Daniel Jacobowitz wrote:
> On Thu, Jun 26, 2008 at 12:08:08AM +0100, Pedro Alves wrote:
> > (The crash will still be there for other targets that
> > don't always register the main thread)
>
> Be nice to get rid of the crash, one way or another.

It will take a while to make all targets consistent,
so although I'd prefer to fix them all, practically speaking, the
assertion should be removed for now, and perhaps leave a comment
there so when we do have all targets consistent, the assert will be
catching a real bug.

The fix seems to me to be a one or two liner,
but, as I've said at gdb@, I don't know if we should output
thread-id=0 or not output thread-id at all, or maybe even
output thread-id=all, as if there are no threads in the list,
this is the special case of there being only one "task" really,
but the target is not registering it.

(OTOH, having an assert at least makes it easy to see
which target needs to be changed.)

> OK.

Thanks, I've checked it in, with an obvious change as attached.  I needed
to include gdbthread.h in monitor.c.

-- 
Pedro Alves