On Mon, Nov 13, 2017 at 12:40:16PM +0900, Junio C Hamano wrote: > As an aside. Over time we accumulated quite a many actions that are > all mutually exclusive by nature. I have a feeling that we might be > better off to move away from this implementation. The only thing > that we are getting from the current one-bit-in-a-flag-word is that > we can name the variable "actions" (instead of "action") to pretend > as if we can be given more than one, and then having to check its > value with HAS_MULTI_BITS(actions) to confuse ourselves. > > Instead, perhaps we should introduce an "enum action" that includes > ACTION_UNSPECIFIED that is the initial value for the "action" > variable, which gets set to ACTION_GET, etc. with OPT_SET_INT(). If > we really care about erroring out when given > > $ git config --add --get foo.bar > > instead of the "last one wins" semantics, we can use OPT_CMDMODE. > > The above is of course outside the scope of this series, and I am > not sure if it should be done as a preparatory or a follow-up > clean-up. Yes, I agree that it's a little confusing, and that an enum is a better representation. The TYPE constants have the same problem. I _think_ we could use OPT_CMDMODE() for those, too. Despite the name, there is nothing in the parse-options error message that would be inappropriate for something that isn't a cmdmode. Though I care a lot less about "--bool --int" reporting an error (instead of last-one-wins) than I do about "--get --set". -Peff
