Re: [PATCH 15/17] cocci: apply the "revision.h" part of "the_repository.pending"

From: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date: Mar 26, 2023, 04:59 UTC
Message-ID: <230326.86ileow1fu.gmgdl@evledraar.gmail.com>
In-Reply-To: <kl6l5yase416.fsf@chooglen-macbookpro.roam.corp.google.com>

On Wed, Mar 22 2023, Glen Choo wrote:

Show 25 quoted lines

> Glen Choo <chooglen@google.com> writes:
>
>> Ævar Arnfjörð Bjarmason         <avarab@gmail.com> writes:
>>
>>> diff --git a/contrib/coccinelle/the_repository.pending.cocci b/contrib/coccinelle/the_repository.pending.cocci
>>> deleted file mode 100644
>>> index 1190a3312bd..00000000000
>>> --- a/contrib/coccinelle/the_repository.pending.cocci
>>> +++ /dev/null
>>> @@ -1,14 +0,0 @@
>>> -// This file is used for the ongoing refactoring of
>>> -// bringing the index or repository struct in all of
>>> -// our code base.
>> We can't go so far as to say that we've removed all implicit references
>> to "the_repository", though, since we still have functions that
>> reference "the_repository" in their implementations. But, I don't think
>> this ".cocci" file would help us with those cases anyway, since this was
>> targeted specifically at functions/macros that were passing
>> "the_repository" to functions that accepted a "struct repository" arg.
>
> For these implicitly-the_repository functions, (e.g. git_path) we'd
> presumably refactor them into repo_* versions and then apply the same
> sorts of changes we did in this series? I guess we'd make those changes
> in contrib/coccinelle/the_repository.cocci, so we don't need the
> *.pending* one.

Whether we add it to a "pending" or not is just a question of whether the migration is done right away, or left for later.

> On that note, I'm curious what contrib/coccinelle/the_repository.cocci
> is doing for us after this series. By definition, all of the macros have
> been fully migrated, so they're all a noop.

I left them for the benefit of any in-flight conflicts, or semantic conflicts with out-of-tree.

I.e. in such a case you'd keep the other side, then apply the cocci rule, and the result is a semantically correct merge of the two.

> Would this slow down `make coccicheck`?

A bit. It doesn't slow it down by much, as these rules are the simplest to execute, spatch can skip the file if the relevant tokens aren't found in it.