From: Patrick Steinhardt Date: Wed, 09 Sep 2026 09:23:43 GMT Subject: Re: [PATCH v2 11/11] setup: allow "git init --ref-storage-format=" to specify a payload Message-ID: In-Reply-To: On Wed, Sep 09, 2026 at 02:24:28PM +0530, Kaartic Sivaraam wrote: > On 9/7/26 16:48, Patrick Steinhardt wrote: > > diff --git a/Documentation/ref-storage-format.adoc b/Documentation/ref-storage-format.adoc > > index c5e29ec831..21d62557b7 100644 > > --- a/Documentation/ref-storage-format.adoc > > +++ b/Documentation/ref-storage-format.adoc > > @@ -1,8 +1,12 @@ > > -`files`;; for loose files with packed-refs. > > +`files[://]`;; for loose files with packed-refs. The optional payload can > > +be specified to change the root directory where references are created. A > > +relative path will be resolved relative to the repository's common directory. > > ifndef::with-breaking-changes[] > > This is the default. > > endif::with-breaking-changes[] > > -`reftable`;; for the reftable format. > > +`reftable[://]`;; for the reftable format. The optional payload can > > +be specified to change the root directory where references are created. A > > +relative path will be resolved relative to the repository's common directory. > > ifdef::with-breaking-changes[] > > This is the default. > > endif::with-breaking-changes[] > > I think we need to be a bit more careful when modifying this ascii doc. This > is also included in Documentation/git-refs.adoc and > Documentation/git-repo.adoc. > > We need to check if support should also be added to the former. For the > latter, I think the inclusion may be fine since it is only for informational > purposes. Yeha, for git-repo(1) it's a good change even, I'd argue. But for the former I'd also argue that git-refs(1) itself should learn to use URIs here. But that's a bigger change, and one that I'd rather avoid doing as part of this already-huge patch series. You know, I'll probably keep this specific change out and only change documentation of git-init(1) and git-clone(1) for now. Thanks! Patrick