[PATCH v2 3/5] Makefile: clean '.c' files generated by carray

[Ivan Orlov]

On 2/28/24 16:32, Andrew Jones wrote:
> On Wed, Feb 28, 2024 at 04:18:28PM +0000, Ivan Orlov wrote:
>> On 2/28/24 14:23, Andrew Jones wrote:
>>> On Thu, Feb 15, 2024 at 04:16:23PM +0000, Ivan Orlov wrote:
>>>> `make clean` doesn't clear the '*.c' files generated by carray.sh
>>>> in the 'build' directory. Fix it by extending the 'clean' makefile
>>>> target.
>>>>
>>>> This is the only way we are able to regenerate the carray .c file in
>>>> case if we add a new test to the 'carray-sbi_unit_tests-y' Makefile
>>>> variable. However, running `make clear` every time we add a new test
>>>> is not really convenient, so the mechanics should be changed in the
>>>> future.
>>>>
>>>> Signed-off-by: Ivan Orlov <ivan.orlov0322 at gmail.com>
>>>> ---
>>>>    Makefile | 2 ++
>>>>    1 file changed, 2 insertions(+)
>>>>
>>>> diff --git a/Makefile b/Makefile
>>>> index 680c19a..4519277 100644
>>>> --- a/Makefile
>>>> +++ b/Makefile
>>>> @@ -684,6 +684,8 @@ clean:
>>>>    	$(CMD_PREFIX)find $(build_dir) -type f -name "*.bin" -exec rm -rf {} +
>>>>    	$(if $(V), @echo " RM        $(build_dir)/*.dtb")
>>>>    	$(CMD_PREFIX)find $(build_dir) -type f -name "*.dtb" -exec rm -rf {} +
>>>> +	$(if $(V), @echo " RM        $(build_dir)/*.c")
>>>> +	$(CMD_PREFIX)find $(build_dir) -type f -name "*.c" -exec rm -rf {} +
>>>
>>> I realize these *.c files are in $build_dir, so they're supposed to be
>>> generated files, but it still looks scary to remove source files. I
>>> think we should try to be more specific. Maybe we should put all generated
>>> source files in directories named 'generated' or something and then only
>>> remove files in those directories.
>>>
>>
>> Hmm, maybe it would be better to do this in the separate patch series? I was
>> thinking about upgrading carrays in general: now the generated .c files
>> don't get updated when we change the corresponding Makefile variable. It
>> produces inconvenience and these variables should be tracked somehow.
> 
> Yes, a separate series is a good idea. For this series, I'd just make the
> new carrays behave like the rest, even if that behavior isn't ideal.
> 

Alright, then I'll drop this patch by now and mention in the docs that 
you may need to manually remove autogenerated .c files in order to make 
the new test work.

Thanks!

-- 
Kind regards,
Ivan Orlov