Re: [tuples] walk_tree for tuples

["Christopher Matthews" <chrismatthews at google dot com>]

depending on whether you want to spread the "shorthand" all around the
system - or just keep it in one spot.

Is the general use that someone would go through all of the labels, or
all but the default?  So far everything that I have seen the default
is a special case - so having a separate accessor for it makes sense.
If all labels are treated equal in most of the rest of the system -
then I think including the default is a good idea.

On 7/12/07, Diego Novillo <dnovillo@google.com> wrote:
> On 7/12/07 2:50 PM, Christopher Matthews wrote:
> > Are you sure that there is a one off error? In gs_switch_set_label I
> > add one to the incoming position to make room for the default.
> >
> >> static inline void
> >> gs_switch_set_label (gimple gs, int index, tree label)
> >> {
> >>  GS_CHECK (gs, GS_SWITCH);
> >>  gcc_assert ((unsigned)index <= gs->gs_switch.nlabels);
> >>  gs->gs_switch.labels[index + 1] = label;
> >> }
> >
> > I did it this way to keep the fact that the default label was stored
> > in [0] away from the build implementation. Or am I completely missing
> > where the 1 off is?
>
> I don't see an off by one error here.  However, the gs_switch_label and
> gs_switch_set_label shouldn't add 1 to INDEX.  It should be possible to
> get at label[0] using them.  The gs_switch_default_label accessor is
> just a clarifying shorthand.