FourCC: Improve debug prints

[dcz-self]

This looks bad:

FourCC {
                repr: [
                    89,
                    85,
                    89,
                    86,
                ],
            },

This looks better:

FourCC("YUYV"),

Fallback for non-ASCII:

FourCC(01 ff 20 cd)

I think drm_fourcc is a better choice for representing fourcc in Linux, but I'm not ready to propose such a change.

[MarijnS95]

I think drm_fourcc is a better choice for representing fourcc in Linux, but I'm not ready to propose such a change.

I had once done a conversion to drm_fourcc but never ended up PR'ing it because that crate, as its name implies, was only ever designed to hold enum variants for the DRM crate, and enum variants cannot be extended externally. Quite a bunch of the V4L2_PIX_FMT_* enums defined at https://elixir.bootlin.com/linux/master/source/include/uapi/linux/videodev2.h are not represented in it currently, and I don't know if we can request the drm_fourcc maintainers to include the V4L2 types. Because of that, DrmFourcc is lossy and any conversion for an unknown fourcc code will result in an UnrecognizedFourcc error/carrier.

[dcz-self]

I just asked dzfranklin/drm-fourcc-rs#25

[MarijnS95]

@dcz-self I think you pushed some of your own testing to the master branch, which was used as a base branch for this PR.

To not have to close this PR and lose all context, perhaps push that to a separate branch :)

[dcz-self]

Oops. Thanks for noticing.

Anyway, is there any interest in this?

[MarijnS95]

I wonder if having the first two bytes always be characters is also true for the FourCC codes used here? Could we "borrow" the formatting logic from drm_fourcc, and/or have the same logic shared with Display instead of not writing anything at all if it isn't a UTF-8 string?

Regarding using drm_fourcc, we don't seem to get a response on that issue and the crate is most likely specific to DRM. There are many other crates defining and using FourCC types (think the million crates out there copy-pasting ISO-BMFF logic), and I doubt anyone would be willing to define a generic FourCC crate and manage to get that used by the majority of the popular crates...

[MarijnS95]

Unlikely whether this rust-lang crate could be reworked and revived... it's pretty much "squatting" that name now. https://crates.io/crates/fourcc