[Numpy-discussion] The "I" dtype character

Nathaniel Smith njs at pobox.com
Fri Apr 12 04:13:20 EDT 2013

On Mon, Apr 8, 2013 at 8:14 AM, Georg Brandl <g.brandl at gmx.net> wrote:
> Hi,
> is it intentional that "I" is supported as a dtype character, but cannot be
> suffixed with a size?
>>>> dtype('i1')
> dtype('int8')
>>>> dtype('I1')
> dtype('uint32')

"i" means "integer". "i1" means "integer with 8 bits".

"I" means "32-bit unsigned integer". "I1" means "32-bit unsigned
integer with 8 bits".

Obviously this last thing doesn't make much sense :-). Historically
numpy as accepted it anyway, and just ignored the suffix. In current
numpy it's still allowed for backwards compatibility, but deprecated,
and will become an error at some point. See


More information about the NumPy-Discussion mailing list