Marc-André Lureau <marcandre.lur...@redhat.com> writes:
> And misc code style fix.
>
> Signed-off-by: Marc-André Lureau <marcandre.lur...@redhat.com>
> ---
> include/qapi/qmp/qlit.h | 8 ++++----
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/include/qapi/qmp/qlit.h b/include/qapi/qmp/qlit.h
> index f36bca4554..1e9696988a 100644
> --- a/include/qapi/qmp/qlit.h
> +++ b/include/qapi/qmp/qlit.h
> @@ -36,13 +36,13 @@ struct QLitDictEntry {
> };
>
> #define QLIT_QNUM(val) \
> - (QLitObject){.type = QTYPE_QNUM, .value.qnum = (val)}
> + { .type = QTYPE_QNUM, .value.qnum = (val) }
> #define QLIT_QSTR(val) \
> - (QLitObject){.type = QTYPE_QSTRING, .value.qstr = (val)}
> + { .type = QTYPE_QSTRING, .value.qstr = (val) }
> #define QLIT_QDICT(val) \
> - (QLitObject){.type = QTYPE_QDICT, .value.qdict = (val)}
> + { .type = QTYPE_QDICT, .value.qdict = (val) }
> #define QLIT_QLIST(val) \
> - (QLitObject){.type = QTYPE_QLIST, .value.qlist = (val)}
> + { .type = QTYPE_QLIST, .value.qlist = (val) }
>
> int compare_litqobj_to_qobj(QLitObject *lhs, QObject *rhs);
Technically, this isn't a type cast, it's a compound literal. A
compound literal is "an unnamed object whose value is given by the
initializer list" (C99 §6.5.2.5). The standard then cautions "that this
differs from a cast expression" (ibid).
The previous paragraph probably makes sense only to language-lawyers, so
let's consider an example.
The macro
#define QLIT_QSTR(val) \
(QLitObject){.type = QTYPE_QSTRING, .value.qstr = (val)}
expands into a compound literal of type QLitObject, which can be used as
an initializer for a variable of type QLitObject.
The macro
#define QLIT_QSTR(val) \
{ .type = QTYPE_QSTRING, .value.qstr = (val) }
expands into something that can be used as an initializer for a variable
of *any* structure type with suitable members. Duck typing, if you
will.
The original author's choice of compound literals is clearly
intentional. I'd prefer to respect it unless there's a compelling
reason not to.
