Re: [PATCH 1/2] ref-filter: hacky "streaming" mode

[Jeff King <peff@peff.net>]

On Thu, Sep 09, 2021 at 10:45:15PM +0800, ZheNing Hu wrote:

> @@ -2735,6 +2723,7 @@ void pretty_print_ref(const char *name, const
> struct object_id *oid,
> 
>         ref_item = new_ref_array_item(name, oid);
>         ref_item->kind = ref_kind_from_refname(name);
> +       read_ref_full(name, 0, NULL, &ref_item->flag);
>         if (format_ref_array_item(ref_item, format, &output, &err))
>                 die("%s", err.buf);
>         fwrite(output.buf, 1, output.len, stdout);

IMHO this is the wrong place to do it, since the caller may already have
the flags (and looking up the ref again is a non-trivial cost).

The caller in builtin/tag.c should switch to using read_ref_full() and
pass in the flags.

The caller in builtin/verify-tag.c _probably_ should resolve the ref in
the same way and pass in that full refname and flags. I do worry that
this may be a compatibility problem, but the current behavior seems so
broken to me.

> >   - I suspect people may be relying on the current behavior. The
> >     original was added to be able to compare the internal tagname to the
> >     refname. I.e., that:
> >
> >       git tag -v --format='%(refname) %(tag)' foo
> >
> >     would show "foo foo". Now that _should_ be "%(refname:strip=2)", I
> >     think, but we'd probably be breaking scripts. OTOH, it really feels
> >     like _not_ handing over a real, fully-qualified refname to the
> >     ref-filter code will mean other things are broken (e.g.,
> >     ATOM_UPSTREAM is assuming we have a fully-qualified ref).
> >
> 
> This is indeed a sad thing: A bug becomes a feature.
> 
> >     I think a backwards-compatible way of fixing it would be to have
> >     this call hand over the full refname to the ref-filter code, but
> >     tell it that %(refname) should default to strip=2. And then anybody
> >     who wants the full name can use %(refname:strip=0).
> >
> 
> Doesn't this make things more complicated? Those callers of git for-each-ref,
> wouldn't our changes like this destroy them?

My proposal was that we'd have a specific flag in ref-filter to say
"default refname:strip to this value". And then _only_ "tag --verify"
would set that (to "2"), leaving for-each-ref, etc unaffected.

So yes, it's complicated. And it must be explained to the user that
"%(refname)" behaves slightly differently with "git tag --verify", but
that is unavoidable if we do not want to break scripts (it _already_
behaves slightly differently, and we just never told anyone).

The other option is to declare the current behavior a bug and fix it. I
am quite tempted by that route, given the inconsistency with other
formatters, including even "git tag --list --format=%(refname)"!

-Peff