rev-parse: -- is sometimes a flag and sometimes an arg?

> ```
> $ git rev-parse unknown-tag
> unknown-tag
> fatal: ambiguous argument 'unknown-tag': unknown revision or path not
> in the working tree.
> Use '--' to separate paths from revisions, like this:
> 'git <command> [<revision>...] -- [<file>...]'
> ```
>
> OK
>
> ```
> $ git rev-parse unknown-tag --
> fatal: bad revision 'unknown-tag'
> ```
>
> Much nicer error.  But:
>
> ```
> $ git rev-parse HEAD --
> 113a928239196d0d9f70671517ce917071ceecf6
> --
> ```
>
> That's not very nice.  Why is "--" treated as an arg?
>
> ```
> $ git rev-parse HEAD
> 113a928239196d0d9f70671517ce917071ceecf6
> ```
>
> Looking at rev-parse code it SEEMS to be intentional, but I can't
> comprehend why.

>
> Tim Hockin <thockin@google.com> writes:
>
> > ```
> > $ git rev-parse unknown-tag
> > unknown-tag
> > fatal: ambiguous argument 'unknown-tag': unknown revision or path not
> > in the working tree.
> > Use '--' to separate paths from revisions, like this:
> > 'git <command> [<revision>...] -- [<file>...]'
> > ```
> >
> > OK
> >
> > ```
> > $ git rev-parse unknown-tag --
> > fatal: bad revision 'unknown-tag'
> > ```
> >
> > Much nicer error.  But:
> >
> > ```
> > $ git rev-parse HEAD --
> > 113a928239196d0d9f70671517ce917071ceecf6
> > --
> > ```
> >
> > That's not very nice.  Why is "--" treated as an arg?
> >
> > ```
> > $ git rev-parse HEAD
> > 113a928239196d0d9f70671517ce917071ceecf6
> > ```
> >
> > Looking at rev-parse code it SEEMS to be intentional, but I can't
> > comprehend why.
>
> Because the first function of "rev-parse" (among many others) is to
> be a helper usable in scripted Porcelains to "parse" command line
> arguments, shifting them into revs and "files".  Without options
> like --revs-only, --no-revs, --flags etc., it dumps everything, while
> translating "revs" into raw object names, to its output.  "--" is a
> valid command line option for whatever scripted Porcelain you'd be
> writing using "rev-parse" to help parsing its command line arguments.
>
> So everything we see above is very much working as designed.
>

>
> Tim Hockin <thockin@google.com> writes:
>
> > Thanks.  Is there a more friendly way to do what I'm trying to
> > express?  Should I just use `rev-list -n1` instead?
>
> That's hard to answer, as you didn't exactly say what you are trying
> to express.  All you asked was: why is "--" treated as an arg.

> Sorry, I assumed it was clear, but that was dumb of me.
>
> I have a string - might be a tag name, might be a branch name, might
> be a SHA (complete or partial).  I want to resolve that into a
> canonical SHA.

> Tim Hockin <thockin@google.com> writes:
> 
> > Sorry, I assumed it was clear, but that was dumb of me.
> >
> > I have a string - might be a tag name, might be a branch name, might
> > be a SHA (complete or partial).  I want to resolve that into a
> > canonical SHA.
> 
> "git rev-parse --verify string" would insist that 'string' is an
> object name and show it as an object name to the standard output, or
> gives an error message and exits with a non-zero status.

>
> On 2022-08-21 at 00:54:40, Junio C Hamano wrote:
> > Tim Hockin <thockin@google.com> writes:
> >
> > > Sorry, I assumed it was clear, but that was dumb of me.
> > >
> > > I have a string - might be a tag name, might be a branch name, might
> > > be a SHA (complete or partial).  I want to resolve that into a
> > > canonical SHA.
> >
> > "git rev-parse --verify string" would insist that 'string' is an
> > object name and show it as an object name to the standard output, or
> > gives an error message and exits with a non-zero status.
>
> Note that this will print a full hex object ID whether it exists in the
> repo or not if one is given (for example, the all-zeros object ID).  If
> you want to verify that the object exists, write this:
>
> git rev-parse --verify string^{object}
> --
> brian m. carlson (he/him or they/them)
> Toronto, Ontario, CA