sensor_msgs/CompressedImage: updated description of format field

[ijnek]

Port of ros/common_msgs#184 to ROS 2.

From original PR:

The description of the format field has long been outdated. This PR fixes it to be up-to-date with the core image_transport plugins.

[fujitatomoya]

lgtm

[fujitatomoya]

@mjcarroll this one comes from ros/common_msgs#184, can you take a look?

[clalancette]

In short, this looks good to me and does seem to correspond to what the code currently supports. I'm going to go ahead and run CI on it, thanks!

[clalancette]

CI:

• Linux
• Linux-aarch64
• Windows

[ijnek]

Thanks @fujitatomoya and @clalancette !

[amacneil]

The description of the format field has long been outdated. This PR fixes it to be up-to-date with the core image_transport plugins.

Is there any info on when this became outdated? Fwiw I'm pretty sure we support only the "simple" values and not the "verbose" ones in Foxglove, and I don't recall any complaints or feature requests about this, so I'm surprised to hear that the simple ones are unused.

If the field is empty or does not correspond to the above pattern, the image is treated as png image.

If I'm reading this change correctly, the old "simple" format values of jpeg and png are now unsupported, and jpeg should be interpreted as png?

[ijnek]

I don't have details on when simple jpeg, png, etc. were dropped (it may still work with compressed subscribers in image_transport_plugins, i'll have to give it a try), but comprssed publisher and compressed depth publisher from image_transport_plugins has been using the new format for at least 12 years (ros-perception/image_transport_plugins@e9c2768).

As a side note, since compressed images and compressed depth images both use CompressedImage, formats like "png" could be ambiguous since both compressed png image and compressed depth png image exist.