On 4/4/21 7:19 AM, Junio C Hamano wrote:
Nothing else needs to be said. The update-main-cache-tree is a thin
wrapper to work on cache tree for THE MAIN in-core index instance,
so it is natural if it needs to pass r and r->index, exactly because
the latter is THE MAIN in-core index instance for the repository.
That is too obvious that it is best left unsaid, for the same reason
why you shouldn't even mention what you did in updating the use of
update_cache_tree() inside unpack_trees(). There is nothing notable
in there.
cf. Documentation/SubmittingPatches
[[describe-changes]]
[[summary-section]]
[[meaningful-message]]
[[imperative-mood]]
Okay, I'll update the message. Thanks for the review.
This commit also fixes the `sparse-index.c` file in which
the `convert_to_sparse()` and `ensure_full_index()`
method use `cache_tree_update()`.
What branch did you base this patch on? There is no update_cache_tree()
call in sparse-index.c on 'master' (well, there is no sparse-index.c
in 'master' to begin with).
Actually the patch hit against a use of `cache_tree_update()` in
the branch 'ds/sparse-index', so I was suggested by Derrick Stolee
to rebase my work on to of his branch.
You may view the discussion here:
https://lore.kernel.org/git/f187df01-8e59-ac74-01e1-586a7a63fd4e@xxxxxxxxx/
And if you are building somewhere other than the tip of 'master', it
is necessary to say where, in order to save confusion and wasted
effort for your readers. E.g. "This patch was made on top of
merging topic 'X' and 'Y' into 'master'". Then you'd only have to
wait for 'X' and 'Y' to graduate to 'master' before the topic can
proceed.
Thanks.
Should this information be added to the commit message
or just the cover letter of the patch?
