Re: [PATCH] Allow matching of empty structs and arrays

[Ori Bernstein <ori@xxxxxxxxxxxxxx>]

On Sun,  9 Jul 2017 01:47:07 -0400, "S. Gilles" <sgilles@xxxxxxxxxxxx> wrote:

> ---
> This is regarding conversation on #myrddin. Please let me know if
> something should be changed in style or substance.
> 
>  mi/match.c            |  8 +++++---
>  test/empty-struct.myr | 26 ++++++++++++++++++++++++++
>  test/tests            |  1 +
>  3 files changed, 32 insertions(+), 3 deletions(-)
>  create mode 100644 test/empty-struct.myr
> 
> diff --git a/mi/match.c b/mi/match.c
> index 29f8cc44..3d159624 100644
> --- a/mi/match.c
> +++ b/mi/match.c
> @@ -246,9 +246,11 @@ static int acceptall(Dtree *t, Dtree *accept)
>  	return ret;
>  }
>  
> -static int isbasictype(Dtree *dt, Type *ty)
> +static int isnonrecursive(Dtree *dt, Type *ty)
>  {
> -	return istyprimitive(ty) || ty->type == Tyvoid || ty->type == Tyfunc || ty->type == Typtr;
> +	return istyprimitive(ty) || ty->type == Tyvoid || ty->type == Tyfunc ||
> +	ty->type == Typtr || (ty->type == Tystruct && ty->nmemb == 0) ||
> +	(ty->type == Tyarray && fold(ty->asize, 1)->expr.args[0]->lit.intval == 0);
>  }

Just a minor stylistic nitpick: I'd break up the condition into a few
smaller cases -- it's getting unweildy to read.

Another minor nit is that wrapped statements get an extra leading tab:

	foo && bar &&
		baz

Not worth another round of back and forth, so I just fixed them when I committed.

-- 
    Ori Bernstein