On Tue, Apr 18, 2017 at 07:40:37PM -0700, Junio C Hamano wrote:
> > It might even be possible to detect the existing line and
> > have parse-options automatically respect "--foo" when "--no-foo" is
> > present. But that may run afoul of callers that add both "--foo" and
> > "--no-foo" manually.
>
> True but wouldn't that something we would want to avoid anyway?
> That is, "git cmd [--OPT | --no-OPT | --no-no-OPT]" from the end
> user's point of view should be an error because it is unclear what
> difference there are between --OPT and --no-no-OPT. And we should
> be able to add a rule to parse_options_check() to catch such an
> error.
I meant that if you have something like this in your options array:
{ 0, "foo", OPTION_INTEGER, &foo, 1 },
{ 0, "no-foo", OPTION_INTEGER, &foo, 2 },
that if we start magically treating "--no-foo" magically, it will
conflict with "--foo" (in this case that's maybe OK because --foo comes
first, but as a general rule it's dangerous to existing options arrays).
> Having said that, I am not sure if we want to go the route of
> "existing line that begins with 'no-' behaves magical". For
> boolean, I suspect we may be get away with such a magic without
> confusing ourselves too much, though.
Yeah, at which point we might as well ask callers to explicitly ask for
the behavior with OPT_NEGBOOL.
-Peff