Stop validating omero data structure (?)

[dstansby]

In version 0.5 of the spec, this block of text has gone:

The "omero" metadata is optional, but if present it MUST contain the field "channels", which is an array of dictionaries describing the channels of the image.
Each dictionary in "channels" MUST contain the field "color", which is a string of 6 hexadecimal digits specifying the color of the channel in RGB format.
Each dictionary in "channels" MUST contain the field "window", which is a dictionary describing the windowing of the channel.
The field "window" MUST contain the fields "min" and "max", which are the minimum and maximum values of the window, respectively.
It MUST also contain the fields "start" and "end", which are the start and end values of the window, respectively.

Replaced with simply:

See the OMERO WebGateway documentation for more information.

Does this mean we should stop validating the omero metadata like we did for 0.4, since it's no longer hard coded into version 0.5 of the OME-zarr spec?

[jluethi]

I agree that it can be read that way.

Given that this hopefully will get replaced with an actual channels definition eventually, it would be a pity if 0.4 has a decently validated Omero that can be used as a channel definition, 0.5 to 0.x has basically no omero info validation and 0.x+1 would introduce new metadata that would be nicely validated again.

[joshmoore]

I think the problem is that #191 was likely missing from the latest version that @normanrz based his work on. Unless someone else confirms first, I'll review the diffs ASAP.

[dstansby]

Given that text doesn't exsit in version 0.5 of the spec, I think we should not validate the omero metadata here. If it returns in the next version of the spec, it will be easy enough for us to put it back.

[normanrz]

It was definitely not my intention to remove it. I don't know how it got removed, but it should be put back in the doc and I think it should be validated the same way as in 0.4. Fwiw, it is still part of the json schema https://github.com/ome/ngff/blob/main/0.5/schemas/image.schema#L75

[dstansby]

I know it seems very strict and harsh, but our intention with this project was always to follow the spec to the word - if words have gone from a versioned release, then we don't want to (and shouldn't) follow them, even if it happened by accident. After all, there's no way for other folks just reading the spec that there's an extra missing bit they should follow.

[dstansby]

I've opened an upstream issue here: ome/ngff#286