Re: [PATCH 1/5] *.h: move some *_INIT to designated initializers
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Jul 1, 2021, 15:56 UTC
- Message-ID
- <xmqq8s2qrps1.fsf@gitster.g>
- In-Reply-To
- <CAN0heSpsg=bwX9gbYfeWYqUbuen-zArR69FWxNmdtcQW_RubzA@mail.gmail.com>
Martin Ågren <martin.agren@gmail.com> writes:
Show 16 quoted lines
>> -#define STRING_LIST_INIT_NODUP { NULL, 0, 0, 0, NULL }
>> +#define STRING_LIST_INIT_NODUP { 0 }
>
> This "NODUP" one is a bit of an odd case though. You don't actually
> change it to use designated initializers, as the patch says. Instead you
> change it in a slightly different way. In a sense, you assign the
> literal zero not to strdup_strings, but to the first field, which is a
> pointer, where I would have expected NULL.
>
> I think there's been some recent-ish "we can assign `{ 0 }` even if the
> first field is a pointer, because it's idiomatic and works out fine"
> even if assigning 0 to a pointer looks a bit odd. So it might not be
> wrong as such, but it doesn't match what the commit message claims, and
> I think it would be more clear to use `{ .strdup_strings = 0 }` here:
> We're explicitly interested in *not* duplicating the strings. It's not
> just some default "let's leave everything as zero/NULL".Well reasoned. I agree that it is not wrong per-se, and the solution to use designated initializer is indeed in line with the theme of the topic.