Re: breakpoint error messages

[Michael Snyder <msnyder@redhat.com>]

Grace Sainsbury wrote:
> 
> How about this?
> 
> The memory_error is only relevant for memory
> breakpoints, and only insert_breakpoint returns an errno (the hardware
> and catcpoint functions return -1 on error)-- so the call to strerror
> prints the error when it is relevant, and so replaces print_sys_errmsg.
> 
> grace


Much better, thanks.  The only concern I still have is about the msg
"the same program may be running in another process".  Your patch will
now print that message much more often than before, both because you
eliminated the #ifdef, and because you're going to print it any time
-any- error or warning occurs.

Even though we all hate ifdefs, I suggest you leave this one in, 
since taking it out will result in a behavior change, and I suggest
you use another boolian variable to determine when it should be
printed (set the variable true at the point where the existing code
prints the message).

Given those changes, you can go ahead and check it in without 
re-submitting.

Thanks,
Michael