# straight from scalar to SIMD will have the exact same
# width at all elwidths, because layout() detects the integer
# case and converts it, preserving the width at all elwidths
- return SimdShape(self, width=None, signed=signed,
- widths_at_elwid=width)
+ # the names are preserved to ensure parameter-compatibility
+ # with Shape()
+ return SimdShape(self, width=width, # actually widths_at_elwid
+ signed=signed,
+ fixed_width=None)