Re: [PATCH v6 12/15] [GSOC] cat-file: reuse ref-filter logic

From: ZheNing Hu <adlternative@gmail.com>
Date: Jun 28, 2021, 13:51 UTC
Message-ID: <CAOLTT8RLX2Zn7C3hd5JhiAYK9TptXbraf=jobL7Cj+p=uLitvA@mail.gmail.com>
In-Reply-To: <CA+CkUQ-gHT=g3KwcBXkp8eBX7wfY_HkT1=hhRiYLap-Av2w5MA@mail.gmail.com>

Hi,

Hariom verma <hariom18599@gmail.com> 于2021年6月28日周一 下午3:47写道：

Show 17 quoted lines

>
> Hi,
>
> >
> > Most of the atoms in `for-each-ref --format` are now supported,
> > such as `%(tree)`, `%(parent)`, `%(author)`, `%(tagger)`, `%(if)`,
> > `%(then)`, `%(else)`, `%(end)`. But these atoms will be rejected:
> > `%(refname)`, `%(symref)`, `%(upstream)`, `%(push)`, `%(worktreepath)`,
> > `%(flag)`, `%(HEAD)`, because our objects don't have a refname.
>
> It's not clear why some atoms are rejected?
>
> Are we going to support them in later commits? (or sometime in the future)
> OR
> We are never going to support them. Because they make no sense to
> cat-file? (or whatever the reason)
>

Because in "git for-each-ref"'s "family", ref_array_item is generated by filter_refs(), which uses ref_filter_handler() to fill ref_array_item with ref's data. In "git cat-file", we care about the object, not the ref. Therefore, ref_array_item is only filled with {oid, rest} in batch_object_write() in cat-file.c. We cannot represent some specific ref-related data in "git cat-file", so we cannot have some atoms in ref_filter. Yes, we probably won't support them in the future.

From an object-oriented point of view, the atom supported by "cat-file" should be a parent class, "for-each-ref"',"branch","tag"... they have more specific object details (ref), their supported atom should be a derived class, they can support more atoms.

> Whatever is the reason, I think it's a good idea to include it in the
> commit message.
>

Yeah. The sentence "because our objects don't have a refname." may not correctly express the reason for rejecting these atoms. I will add more descriptions.

> --
> Hariom

Thanks. -- ZheNing Hu