On Mon, Feb 3, 2025 at 11:39 PM David Lechner [off-list ref] wrote:

On 2/1/25 1:47 PM, Andy Shevchenko wrote:

quoted

On Sat, Feb 1, 2025 at 6:22 PM Bartosz Golaszewski [off-list ref] wrote:

quoted

On Sat, Feb 1, 2025 at 5:17 PM David Lechner [off-list ref] wrote:

quoted

On 2/1/25 10:14 AM, Bartosz Golaszewski wrote:

quoted

On Sat, Feb 1, 2025 at 5:09 PM David Lechner [off-list ref] wrote:

quoted

On 2/1/25 4:36 AM, Bartosz Golaszewski wrote:

...

quoted

quoted

quoted

quoted

quoted

This looks good to me except for one thing: the function prefix. I would
really appreciate it if we could stay within the existing gpiod_ namespace and
not add a new one in the form of gpiods_.

Maybe: gpiod_multiple_set_ or gpiod_collected_set...?

I was waiting for someone to complain about the naming. ;-)

I was going for as short as possible, but OK, the most obvious prefix to me
would be `gpio_descs_...` (to match the first parameter). Any objections to
that?

Yes, objection! As far as any exported interfaces go: in my book
"gpio_" is the prefix for legacy symbols we want to go away and
"gpiod_" is the prefix for current, descriptor-based API. Anything
else is a no-go. I prefer a longer name that starts with gpiod_ over
anything that's shorter but doesn't.

Oops, that was a typo. I meant to write gpiod_descs_.

Eh... the D in gpioD already stands for "GPIO Descriptor" but if
there's no better option in your opinion than I guess I can live with
that.

gpiod_set_many_value_cansleep() ?

OK, taking all these suggestions into consideration along with having recently
come across regmap_multi_reg_write(), I think I'll go with:

gpiod_multi_set_value_cansleep()

Sounds good.

Bart