On Mon, Feb 21 2022, Abhradeep Chakraborty wrote:
> Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx> wrote:
>
>> Sorry about leaving this patch submission hanging. I read this at the
>> time, but forgot to find time to loop back to it.
>
> No worries. Thanks for reviewing :)
>
>> But in this case there's really a much easier way to do this, to just
>> extend something like this:
>> ...
>> See b6c2a0d45d4 (parse-options: make sure argh string does not have SP
>> or _, 2014-03-23) for the existing code shown in the context where we
>> already check "argh" like that, i.e. we're just missing a test for
>> "help".
>>
>> Obviously such a function would need to hardcode some of the logic you
>> added in your shellscript. E.g. this fires on a string ending in "...",
>> but yours doesn't.
>
> Thank you so much for the suggestion. Didn't aware of it before. I will
> try to implement the logic in parse-options.c` (as you suggested).
>
>> That should be fairly easy to do though, and if not we could always just
>> dump these to stderr or something if a
>> git_env_bool("GIT_TEST_PARSE_OPTIONS_DUMP_FIELD_HELP", 0) was true, and
>> do the testing itself in t0012-help.sh.
>
> Okay but if the logic can't be implented in the `parse-options.c` file
> (most probably I will be able to implement the logic), would you allow me
> to try the `coccinelle script` method you mentioned?
In this case I think there's definitely no reason for why it can't/won't
work in parse-options.c.
If you're doing something like that with coccicheck I'm afraid I can't
help much, I've only seen that the kernel is doing it (it's referenced
in some of the coccinelle docs), but I haven't personally used it for
anything close to that.
