Aarni Koskela

Rebased (post #1773 merge).

(I'll take care of the lint error if there's some review activity on this 👍)

> As discussed in the issue and other PR, this is a breaking change for web service users. Yes, in semver terms this would be a major version bump. (In...

Rebased on 5.0.1.

Rebased. Looking at the changelog for 5.0.0, this could easily be just another "BREAKING" change for 6.0.0?

In my opinion, asking users to change a `geoip2` requirement to `geoip2[requests]` or `geoip2[aiohttp]` depending on what they're doing is not very disruptive 😄 For me and the company I...

Thank you for the response and thoughts. > One thing I do not see though in your code is setting the MCU's processing speed/clocks. I don't know how ESP32 inits...

> I'm also curious why 64 byte chunks are necessary? That seems to be [this bit of code](https://github.com/therealprof/display-interface/blob/8fca041b0288740678f16c1d05cce21bd3867ee5/spi/src/lib.rs#L81-L103). Looks like the intent is to read a 64-byte chunk from the...

> Could you try changing fill_congiguous into a fill_solid, or possibly calling set_pixels directly? I just want to make sure it's not something funky. I changed the draw bit to...

@rfuest Nice work! Am I understanding correctly that since the `Framebuffer` is marked `BigEndian`, the price for converting from native (little-endian) pixel formats to big-endian is paid when drawing to...