[PATCH] Outwit compiler dead code elimination in break.exp test
Mon Dec 1 15:46:00 GMT 2003
Fred Fish writes:
> > > Emit bp_locationNN markers in pass
> > > and fail messages instead of line numbers.
> > Why this? The line numbers were not really hardcoded.
> Generally I believe you should not emit values in the pass/fail
> messages that are subject to change, such as addresses, line numbers,
> etc. This way you can diff successive summary files without seeing
> spurious differences.
> However I can revert that part of the patch if that is an issue.
If the pass/fil messages are unique within the .exp files, then ok.
More information about the Gdb-patches