Re: Bug/regression report - 'git stash push -u' fatal errors when sub-repo present

On 10/4/2021 4:52 PM, Junio C Hamano wrote:

Derrick Stolee [off-list ref] writes:

quoted

And yes, I believe that make_cache_entry() and add_index_entry_with_check()
are the only places that need this internal version. If we find others,
then we can add them as necessary. The tests in t1092 are getting rather
robust, although they don't do much for this test case:

quoted

+test_expect_success 'stash -u ignores sub-repository' '
+	test_when_finished "rm -rf sub-repo" &&
+	git init sub-repo &&
+	git stash -u
+'

Seems like a good test to have, anyway.

I look forward to seeing this as a full patch.

Just one thing I want to pick your brains for ;-)

I said this earlier ...

quoted

quoted

quoted

"git update-index" should never allow to create a "tree" kind of
cache entry (making it sparse again should be the task of systems
internal, and should not be done by end-user feeding a pre-shrunk
"tree" kind of entry to record a sparsely populated state, if I
understand correctly), so its call to verify_path(), if it ends with
a directory separator, should say "that's not a good path".

... without knowing what you had in mind when you did the "tree kind
of entry in the index".  Are we on the same page, or do we think it
might be beneficial to give end-users a long-enough rope
to hang themselves, aka get into the lower details of
implementation?

One _could_ imagine that allowing

 $ git update-index --cacheinfo 40000,609869396314577e5a,t/

given by the end user to drop all entries under t/* and replace them
with a single sparse-dir-entry might be a good way to allow
scripters the same power as the C-code to take advantage of the
sparse checkout feature.  It needs to be paired with some mechanism
to allow sparse-dir-entry observed by the end users with a plumbing,
e.g. even though ls-files unconditionally calls ensure_full_index(),

I think this is an interesting capability, but I'm not sure I see
a use case that is worth the footgun, especially with the accidents
that can happen when working with submodules.

I think we can happily extend to include this functionality in the
future. Having an error condition now that we relax in the future is
the good kind of behavior change.

 $ git ls-files --show-sparse

may show the sparse-dir-entry by bypassing the call.

I have something in the works for this, but I'm letting others send
their sparse-index work first. I have not forgotten your request for
such a feature!

Thanks,
-Stolee