Web lists-archives.com

Re: [PATCH 003/194] object-store: move packed_git and packed_git_mru to object store




[removed rene.scharfe@xxxxxxxxxxxxxx from cc:; I lost that domain a few
 years ago.  Thanks for the heads-up, Stefan!]

Am 12.02.2018 um 20:00 schrieb Stefan Beller:
> On Fri, Feb 9, 2018 at 2:09 PM, Junio C Hamano <gitster@xxxxxxxxx> wrote:
>> Stefan Beller <sbeller@xxxxxxxxxx> writes:
>>
>>> Patch generated by
>>>
>>>   2. Applying the semantic patch contrib/coccinelle/packed_git.cocci
>>>      to adjust callers.
>>
>> About this part...
>>
>>> diff --git a/contrib/coccinelle/packed_git.cocci b/contrib/coccinelle/packed_git.cocci
>>> new file mode 100644
>>> index 0000000000..da317a51a9
>>> --- /dev/null
>>> +++ b/contrib/coccinelle/packed_git.cocci
>>> @@ -0,0 +1,7 @@
>>> +@@ @@
>>> +- packed_git
>>> ++ the_repository->objects.packed_git
>>> +
>>> +@@ @@
>>> +- packed_git_mru
>>> ++ the_repository->objects.packed_git_mru
>>
>> The above is correct for one-time transition turning pre-transition
>> code to post the_repository world, but I am not sure if we want to
>> have it in contrib/coccinelle/, where "make coccicheck" looks at, as
>> a way to continuously keep an eye on "style" violations like using
>> strbuf_addf() for a constant when strbuf_addstr() suffices.
>>
>> Wouldn't we need a mechanism to ensure that this file will *not* be
>> used in "make coccicheck" somehow?

object_id.cocci is similar in this regard -- once the conversion is
done, we won't need it anymore.

> I can omit the cocci files from the patches, if that is better for maintenance.
> 
> I thought it may be a helpful
> for merging this series with the rest of the evolved code base which
> may make use of one of the converted functions. So instead of fixing
> that new instance manually, cocinelle could do that instead.

Right, merging should be easier -- instead of fixing conflicts manually,
Coccinelle could regenerate the patch.

René