[libgpiod] How stable is the v2 API?

[libgpiod] How stable is the v2 API?

[Ben Hutchings]

I'm working on a project that requires GPIO debouncing and is planned
to use libgpiod.  But debouncing is only available on the unreleased
next/libgpiod-2.0 branch.

What we'd like to know is whether the current C API on that branch is
expected to change much before release.  I don't expect any commitment
to API or ABI stability, but it would be helpful to have some
indication of how much change is likely to be needed in a client that
is written for the current API.

Thanks,

Ben.

Re: [libgpiod] How stable is the v2 API?

[Andy Shevchenko]

On Tue, Jul 13, 2021 at 12:09 AM Ben Hutchings
<ben.hutchings@essensium.com> wrote:
>
> I'm working on a project that requires GPIO debouncing and is planned
> to use libgpiod.  But debouncing is only available on the unreleased
> next/libgpiod-2.0 branch.
>
> What we'd like to know is whether the current C API on that branch is
> expected to change much before release.  I don't expect any commitment
> to API or ABI stability, but it would be helpful to have some
> indication of how much change is likely to be needed in a client that
> is written for the current API.

I believe this is the right thread to discuss and influence if needed.

https://lore.kernel.org/linux-gpio/20210621191830.10628-1-brgl@bgdev.pl/T/#ma70b1ab80804c09644f4a018ff5eff10464b195a

-- 
With Best Regards,
Andy Shevchenko

Re: [libgpiod] How stable is the v2 API?

[Andy Shevchenko]

On Tue, Jul 13, 2021 at 12:16 PM Andy Shevchenko
<andy.shevchenko@gmail.com> wrote:
>
> On Tue, Jul 13, 2021 at 12:09 AM Ben Hutchings
> <ben.hutchings@essensium.com> wrote:
> >
> > I'm working on a project that requires GPIO debouncing and is planned
> > to use libgpiod.  But debouncing is only available on the unreleased
> > next/libgpiod-2.0 branch.
> >
> > What we'd like to know is whether the current C API on that branch is
> > expected to change much before release.  I don't expect any commitment
> > to API or ABI stability, but it would be helpful to have some
> > indication of how much change is likely to be needed in a client that
> > is written for the current API.
>
> I believe this is the right thread to discuss and influence if needed.
>
> https://lore.kernel.org/linux-gpio/20210621191830.10628-1-brgl@bgdev.pl/T/#ma70b1ab80804c09644f4a018ff5eff10464b195a

Oops, it was a follow up, but the core bindings were here:
https://lore.kernel.org/linux-gpio/20210607045950.GA18318@sol/T/#u

-- 
With Best Regards,
Andy Shevchenko

Re: [libgpiod] How stable is the v2 API?

[Bartosz Golaszewski]

On Mon, Jul 12, 2021 at 11:08 PM Ben Hutchings
<ben.hutchings@essensium.com> wrote:
>
> I'm working on a project that requires GPIO debouncing and is planned
> to use libgpiod.  But debouncing is only available on the unreleased
> next/libgpiod-2.0 branch.
>
> What we'd like to know is whether the current C API on that branch is
> expected to change much before release.  I don't expect any commitment
> to API or ABI stability, but it would be helpful to have some
> indication of how much change is likely to be needed in a client that
> is written for the current API.
>
> Thanks,
>
> Ben.
>

Hi Ben!

While - as you already assumed - there are no commitments to any API &
ABI stability yet, the interface should not change very much. I have a
single set of API changes in my queue that we discussed while
reviewing the C++ bindings and I don't expect there to be many more
coming after that.

Again: this is an estimation, not a promise.

Best Regards
Bartosz

Re: [libgpiod] How stable is the v2 API?

[Ben Hutchings]

On Tue, Jul 13, 2021 at 12:16:52PM +0300, Andy Shevchenko wrote:
> On Tue, Jul 13, 2021 at 12:09 AM Ben Hutchings
> <ben.hutchings@essensium.com> wrote:
> >
> > I'm working on a project that requires GPIO debouncing and is planned
> > to use libgpiod.  But debouncing is only available on the unreleased
> > next/libgpiod-2.0 branch.
> >
> > What we'd like to know is whether the current C API on that branch is
> > expected to change much before release.  I don't expect any commitment
> > to API or ABI stability, but it would be helpful to have some
> > indication of how much change is likely to be needed in a client that
> > is written for the current API.
> 
> I believe this is the right thread to discuss and influence if needed.
> 
> https://lore.kernel.org/linux-gpio/20210621191830.10628-1-brgl@bgdev.pl/T/#ma70b1ab80804c09644f4a018ff5eff10464b195a

Thanks.  I've actually seen that already.  I don't think I need to
exert any "influence" as the current v2 API seems to do cover our
needs.

Ben.

Re: [libgpiod] How stable is the v2 API?

[Ben Hutchings]

On Tue, Jul 13, 2021 at 10:10:12PM +0200, Bartosz Golaszewski wrote:
[...]
> While - as you already assumed - there are no commitments to any API &
> ABI stability yet, the interface should not change very much. I have a
> single set of API changes in my queue that we discussed while
> reviewing the C++ bindings and I don't expect there to be many more
> coming after that.

Is that single set of API changes the same as
<https://patchwork.ozlabs.org/project/linux-gpio/list/?series=250005>
or do you have more beyond that?

> Again: this is an estimation, not a promise.

Understood.

Thanks,

Ben.

Re: [libgpiod] How stable is the v2 API?

[Bartosz Golaszewski]

On Tue, Jul 13, 2021 at 11:46 PM Ben Hutchings
<ben.hutchings@essensium.com> wrote:
>
> On Tue, Jul 13, 2021 at 10:10:12PM +0200, Bartosz Golaszewski wrote:
> [...]
> > While - as you already assumed - there are no commitments to any API &
> > ABI stability yet, the interface should not change very much. I have a
> > single set of API changes in my queue that we discussed while
> > reviewing the C++ bindings and I don't expect there to be many more
> > coming after that.
>
> Is that single set of API changes the same as
> <https://patchwork.ozlabs.org/project/linux-gpio/list/?series=250005>
> or do you have more beyond that?
>

These have already been squashed with the top commit in the
next/libgpiod-2.0 branch (except for the last one). I have two more
that I will try to send out today. I'll Cc you.

Bart

> > Again: this is an estimation, not a promise.
>
> Understood.
>
> Thanks,
>
> Ben.