Re: [PATCH] for_each_string_list_item(): behave correctly for empty list

From: SZEDER Gábor <szeder.dev@gmail.com>
Date: Sep 19, 2017, 13:45 UTC
Message-ID: <CAM0VKjmuyp0pn29s8=diCpA0FPjjFJH-LcxQmOJH6Z-7WZNurA@mail.gmail.com>
In-Reply-To: <CAM0VKjn=KjTHBoubJKbxx7MasJ6wWcUFrCwrvr5oHwUCsfr_Pw@mail.gmail.com>

On Tue, Sep 19, 2017 at 3:38 PM, SZEDER Gábor <szeder.dev@gmail.com> wrote:

Show 12 quoted lines

> A bit "futuristic" option along these lines could be something like
> this, using a scoped loop variable in the outer loop to ensure that
> it's executed at most once:
>
>   #define for_each_string_list_item(item,list) \
>       for (int f_e_s_l_i = 1; (list)->items && f_e_s_l_i; f_e_s_l_i = 0) \
>           for (item = (list)->items; item < (list)->items + (list)->nr; ++item)
>
> The high number of underscores are an attempt to make reasonably sure
> that the macro's loop variable doesn't shadow any variable in its
> callers or isn't being shadowed in the loop body, which might(?)
> trigger warnings in some compilers.

Well, and a poor attempt at that, because, of course, the loop variable would still be shadowed in nested for_each_string_list_item loops... And our codebase has these loops nested in entry.c:finish_delayed_checkout().

OTOH, we don't seem to care too much about shadowed variables, since building with -Wshadow gives 91 warnings in current master...