On Thu, 16 Aug 2012 13:49:15 -0300 Luiz Capitulino <lcapitul...@redhat.com> wrote:
> > I converted more error messages to the error-reporting-infrastructure- > > du-jour than was enjoyable, and I can tell you that the restrictions > > that come with error_report() compared to anything-goes-fprintf() do > > make the job easier. > > This patch is fixing a function that's only used in command-line context, > I don't see why fprintf() shouldn't be a good fit. Your call about PROGNAME > is a valid one, but no function in the call chain prints it yet, so it's not > a big deal, specially if compared to the alternative (which is using > error_report()). I've talked with Markus about this on IRC and (correct me if I'm wrong Markus), he says that we could kill the HMP stuff from error_report() once no caller is depending on it. Looks like a plan. I'd rather not add new error_report() calls where it's not strictly needed, but no biggie.