On Tue, Dec 11, 2018 at 10:55:33PM +0000, Ramsay Jones wrote:
>
>
> On 09/12/2018 23:43, Luc Van Oostenryck wrote:
> > This allows to have a single function to output the
> > size, the type, the maximal value, ...
> > ---
> > lib.c | 57 ++++++++++++++++++++++++++++++++++++---------------------
> > 1 file changed, 36 insertions(+), 21 deletions(-)
> >
> > diff --git a/lib.c b/lib.c
> > index db33e58b2..1156f6f6c 100644
> > --- a/lib.c
> > +++ b/lib.c
> > @@ -1144,6 +1144,15 @@ static char **handle_switch(char *arg, char **next)
> > return next;
> > }
> >
> > +#define PTYPE_SIZEOF (1U << 0)
> > +#define PTYPE_T (1U << 1)
> > +#define PTYPE_SIZEOF_T (PTYPE_SIZEOF|PTYPE_T)
> > +#define PTYPE_MAX (1U << 2)
> > +#define PTYPE_MIN (1U << 3)
> > +#define PTYPE_TYPE (1U << 4)
> > +#define PTYPE_WIDTH (1U << 4)
>
> (1U << 4) used twice for _TYPE and _WIDTH.
Uh, indeed. Thanks for noticing this.
> > + if (flags & PTYPE_MAX) {
> > + const char *suffix = builtin_type_suffix(type);
> > + predefined_max(name, suffix, bits);
>
> predefined_max() does not account for signed-ness of the type, so
> that _all_ unsigned types have a _MAX value that is the same as
> the signed variant. e.g. __UINT16_MAX__ is 0x7fff rather than 0xffff.
Indeed, when I started this series all the types were signed.
This is fixed now. Thanks!
> > + }
> > + if (flags & PTYPE_TYPE)
> > + predefined_type(name, type);
>
> The type names are not expressed the same as gcc (not a big deal), but
> we could change the type names in the typenames[] table to match.
> (show-parse.c, ln 222).
Yes, I know. In fact it's more or less on purpose as 'unsigned long'
talks to me much more than 'long unsigned int'.
I suppose you only care about a one-to-one textual correspondance
when testing/comparing with what GCC outputs?
Best regards
-- Luc
