Re: [libgpiod][PATCH 8/8] bindings: rust: examples: replace tools examples with use case examples

[Kent Gibson <warthog618@gmail.com>]

On Fri, Jun 23, 2023 at 09:57:17AM +0200, Erik Schilling wrote:
> On Fri Jun 23, 2023 at 6:39 AM CEST, Kent Gibson wrote:
> > Replace tool examples with use case examples drawn from the tools,
> > gpio_events example with buffered_event_lifetimes, and
> > gpio_threaded_info_events with reconfigure_input_to_output.
> >
> > Signed-off-by: Kent Gibson <warthog618@gmail.com>
> 
> Reviewed-by: Erik Schilling <erik.schilling@linaro.org>
> 
> >     let mut buffer = libgpiod::request::Buffer::new(4)?;
> 
> Slightly tangential:
> 
> Maybe the API should provide some sensible defaults for the buffer size?
> (Or just set a sane default and provide ways to override it)? The change
> from 1 -> 4 for bulk operations seems reasonable, but I feel like a user
> just getting started with all of this likely won't know what might be
> good values to pick...
> 


The C API does that - it defaults to 64, IIRC - if you pass in 0.
And the Rust will do the same - read the docs:

    /// Create a new edge event buffer.
    ///
    /// If capacity equals 0, it will be set to a default value of 64. If
    /// capacity is larger than 1024, it will be limited to 1024.

Using 64 seems excessive to me, so I explictly set more reasonable
sizes in the examples.

Btw a casual user is probably fine with 1 - the events will still be
buffered in the kernel so the only advantages of > 1 is to reduce the
number of reads when handling bursts, and so drain the kernel buffer
slightly quicker.  I expect that the only users who would see an
appreciable difference in behaviour with different userspace buffer
sizes probably have a good idea why they want to be setting it.
Again, would be good to document that - if it isn't already.

Cheers,
Kent.