On 9/10/25 18:08, Patrick Steinhardt wrote:
Hi,
On Tue, Aug 26, 2025 at 06:34:02AM +0700, Bagas Sanjaya wrote:
Hi,
Askar Safin <safinaskar@xxxxxxxxxxxx> posted on the ML about his complaints
regarding grafts removal in v3.0+ (see [1] for the full thread). He also
briefly mentioned future of shallow clones (aka git clone --depth=1).
The question is: If grafts support is removed, then how can shallow clones
be implemented in terms of git-replace(1)? What about its implementation
details? How about unshallowing (git fetch --unshallow)?
Thanks.
[1]: https://lore.kernel.org/git/198c29e9058.119e3a5c065010.5888624019176274871@xxxxxxxxxxxx/T/#t
I think the BreakingChanges document could use a revamp in that regard.
The mechanism used to implement shallow clones is of course _not_ going
away, and the internal mechanism to facilitate this will go away.
So the intention really only to remove ".git/info/grafts" as a
user-visible mechanism. Users should exclusively use git-replace(1).
OK, thanks!
--
An old man doll... just what I always wanted! - Clara
