[m-dev.] for review: new method of handling failures, part 7 of 6 :-(

[Zoltan Somogyi]

> > +** This test allows Mercury developers who have not done a cvs update
> > +** of the changes introducing the new failure handling method in the
> > +** code generator to retain the ability to do bootchecks, by adding
> > +** -DMR_DISABLE_REDOFR to EXTRA_CFLAGS in their Mmake.stage.params,
> > +** and using bootcheck -r.
> > +**
> > +** The test should be deleted once there are no such developers.
> > +*/
> > +
> > +#ifdef	MR_DISABLE_REDOFR
> > +#undef	MR_USE_REDOFR
> > +#endif
> 
> s/#undef/  #undef/
> 
> I don't quite understand this code --
> where will MR_USE_REDOFR get #defined in the first place,
> thus requiring that we #undef it here?

As I said in my previous message, eventually it will defined in mercury_imp.h
a couple of lines above the quoted stuff.

> > +	} else if (streq(words[0], "toggle_echo")) {
> > +		MR_supply_newline = !MR_supply_newline;
> 
> Does toggle_echo actually cause the commands to be
> echoed, or just it just result in the printing of
> a newline?  If the latter, then the command name should
> be changed.  If the former, then this code looks wrong.

The variable was misnamed; it is now MR_echo_commands.

> Apparently you haven't done a `cvs add' for cut_test.m
> and cut_test.exp?  At any rate, they didn't appear in the diff.

This is what Tyson's makediff script does.

I have compiled with your other suggestions on this review fragment.

Zoltan.