Add support for EPD 3in7

[mangelajo]

Matches implementation from waveshare/e-Paper code

https://github.com/waveshare/e-Paper/blob/master/RaspberryPi_JetsonNano/c/lib/e-Paper/EPD_3in7.c

Co-Authored-By: Miguel Angel Ajo [email protected]

[mangelajo]

ok, it's ignoring #![allow(non_camel_case_types)]

I guess I will have to change the enum names.

[mangelajo]

@caemor is there any driver doing grayscale or shall I define a new color type?

[mangelajo]

Ah, there is Gray4! https://docs.rs/embedded-graphics/latest/embedded_graphics/pixelcolor/struct.Gray4.html

[peckpeck]

I'd suggest to make the Color a generic parameter and vary it in impl, this way you could factor some of he code

[mangelajo]

I'd suggest to make the Color a generic parameter and vary it in impl, this way you could factor some of he code

Yes, I was thinking about that, but there are a couple of things holding me back:

• The code paths and initialization for both modes (see Gray1/Gray4 functions here [1]) are wildly different, they only seem to share SPI bus access.
• It would special-case this driver in terms of instantiation, I guess we could may be have a default instance in one of the modes -not sure if it's possible, new to rust here-, for other drivers (tri-color vs bw, the drivers are duplicated)

[1] https://github.com/waveshare/e-Paper/blob/master/RaspberryPi_JetsonNano/c/lib/e-Paper/EPD_3in7.c#L171

[mangelajo]

Oh, I will go and implement partial updates for this driver in a separate PR first, while I keep rumiating about the Grayscales ;)

[peckpeck]

If you use a generic type for color, you can easily separate the code for init with an impl for Color and and impl for Gray4. But update_frame will still be difficult, maybe smaller trait would help implementing this kind of cases.

[peckpeck]

If there is no real improvement besides memory consumption, why have a separate BW initialization ? A BW update_frame on a gray initialized device should work no ?

[mangelajo]

If there is no real improvement besides memory consumption, why have a separate BW initialization ? A BW update_frame on a gray initialized device should work no ?

I really have no numbers to compare yet. I will eventually be working on Gray4 support on a different PR. At that point we can analyze it from both the timing and memory perspective and decide.

But it sounds like a good point, if they were to have the same refresh rates, why keep two code paths :)

[mangelajo]

Note: I may need some time to get into Gray4, currently I am working on the application that uses the implementation of this PR.

Hopefully we can merge this if everything looks good, and then treat Gray4 support as an upgrade PR.

[peckpeck]

Indeed, since we in 0.x releases we should merge things quickly to allow for easier testing.

[mangelajo]

I am not sure what's going on, I see the errors here:

https://github.com/caemor/epd-waveshare/actions/runs/3608595550/jobs/6081293541#step:5:105

But with my architecture, it builds all just fine.

Any advice?

[peckpeck]

I am not sure what's going on, I see the errors here:

You need to rebase your code, the main branch has been updated with new commits that impact your code.
Something like git rebase -i origin/main

[mangelajo]

I am not sure what's going on, I see the errors here:

You need to rebase your code, the main branch has been updated with new commits that impact your code. Something like git rebase -i origin/main

Ack, I will rebase and let's see how it goes, I would have expected the unit testing/etc to pass even if it needed rebase, of course github may complain if it's unable to merge your changes.

[mangelajo]

Ok, I rebased and squashed the commits, now at least I get it reproduced locally:

error[E0407]: method `is_busy` is not a member of trait `WaveshareDisplay`
   --> src/epd3in7/mod.rs:260:5
    |
260 | /     fn is_busy(&self) -> bool {
261 | |         self.interface.is_busy(IS_BUSY_LOW)
262 | |     }
    | |_____^ not a member of trait `WaveshareDisplay`

error[E0277]: the trait bound `DELAY: _embedded_hal_blocking_delay_DelayUs<u32>` is not satisfied
  --> src/epd3in7/mod.rs:50:37
   |
50 | impl<SPI, CS, BUSY, DC, RST, DELAY> InternalWiAdditions<SPI, CS, BUSY, DC, RST, DELAY>
   |                                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ the trait `_embedded_hal_blocking_delay_DelayUs<u32>` is not implemented for `DELAY`
   |
note: required by a bound in `traits::InternalWiAdditions`
  --> src/traits.rs:36:12
   |
29 | pub(crate) trait InternalWiAdditions<SPI, CS, BUSY, DC, RST, DELAY>
   |                  ------------------- required by a bound in this
...
36 |     DELAY: DelayUs<u32>,
   |            ^^^^^^^^^^^^ required by this bound in `traits::InternalWiAdditions`
help: consider further restricting this bound
   |
58 |     DELAY: DelayMs<u8> + embedded_hal::prelude::_embedded_hal_blocking_delay_DelayUs<u32>,
   |                        ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

error[E0277]: the trait bound `DELAY: _embedded_hal_blocking_delay_DelayUs<u32>` is not satisfied
   --> src/epd3in7/mod.rs:125:37
    |
125 | impl<SPI, CS, BUSY, DC, RST, DELAY> WaveshareDisplay<SPI, CS, BUSY, DC, RST, DELAY>
    |                                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ the trait `_embedded_hal_blocking_delay_DelayUs<u32>` is not implemented for `DELAY`
    |
note: required by a bound in `traits::WaveshareDisplay`
   --> src/traits.rs:146:12
    |
139 | pub trait WaveshareDisplay<SPI, CS, BUSY, DC, RST, DELAY>
    |           ---------------- required by a bound in this
...
146 |     DELAY: DelayUs<u32>,
    |            ^^^^^^^^^^^^ required by this bound in `traits::WaveshareDisplay`
help: consider further restricting this bound
    |
133 |     DELAY: DelayMs<u8> + embedded_hal::prelude::_embedded_hal_blocking_delay_DelayUs<u32>,
    |                        ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

I will try to have it fixed by tomorrow.

[caemor]

If it helps:
#125 and #131 should be the issues:

• Replace DelayMs with DelayUs
• And replace is_busy with wait_until_idle in the Waveshare Trait Implementation:

[mangelajo]

If it helps: #125 and #131 should be the issues:

* Replace DelayMs with DelayUs

* And replace is_busy with wait_until_idle in the Waveshare Trait Implementation: ![image](https://user-images.githubusercontent.com/11088935/206298426-4e76ed07-19e5-4985-9b30-0d5f7b127a1e.png)

Thanks a lot, it helps, yes! #59

[mangelajo]

hopefully compile now after rebase :-)

[caemor]

Thanks for your great work!

[mangelajo]

Thanks for your great work!

Thank you!! 🙏🏻