On Tue, Dec 12, 2023 at 11:30:22AM -0800, Junio C Hamano wrote:
> Patrick Steinhardt <ps@xxxxxx> writes:
>
> > To accomodate for cases where the host system has no Git installation we
> > use the locally-compiled version of Git. This can result in problems
> > though when the Git project's repository is using extensions that the
> > locally-compiled version of Git doesn't understand. It will refuse to
> > run and thus cause the checks to fail.
> >
> > Fix this issue by prefering the host's Git resolved via PATH. If it
> > doesn't exist, then we fall back to the locally-compiled Git version and
> > diff as before.
> >
> > Signed-off-by: Patrick Steinhardt <ps@xxxxxx>
> > ---
> >
> > I've started to dogfood the reftable backend on my local machine and
> > have converted many repositories to use the reftable backend. This
> > surfaced the described issue because the repository now sets up the
> > "extensions.refStorage" extension, and thus "check-chainlint" fails
> > depending on which versions of Git I'm trying to compile and test.
>
> I do not think "prefer host Git" is necessarily a good idea; falling
> back to use host Git is perfectly fine, of course.
Why is that, though? We already use host Git in other parts of our build
infra, and the options we pass to git-diff(1) have been around for ages:
- `--no-pager` was introduced in 463a849d00 (Add and document a global
--no-pager option for git., 2007-08-19).
- `--no-index` is around since at least fcfa33ec90 (diff: make more
cases implicit --no-index, 2007-02-25).
- `-w` is around since 0d21efa51c (Teach diff about -b and -w flags,
2006-06-14).
So all options have pretty much been there since forever. Which means
that if the host has Git around, and the Git version is at least v1.5.3,
then it also understands all options.
Furthermore, we are accessing the Git repository that the user has set
up with host Git in the first place, so I'd think even conceptually it
is the correct thing to do.
Patrick
Attachment:
signature.asc
Description: PGP signature
