RFC-6: Comment 1#
Summary#
We believe that the RFC should be accepted in its current form, ideally to coincide with the release of RFC-5.
Significant comments and questions#
Additional features tend towards increased complexity, so any changes which can be made to decrease needless complexity are welcome. Zarr (and therefore OME-Zarr) already has a way of representing multiple unrelated groups of datasets: groups.
However, we should respect that this is a breaking change and so try to minimise impact on implementers. For that reason, its release should be timed to coincide with that of RFC-5, which also targets the multiscale specification and is the only other backwards-incompatible RFC currently on the table. RFC-6 probably doesn’t add enough value to be worth the breaking change on its own, but is a nice step towards simplicity as part of another (much more complex) breaking change.
Unlike RFC-5, a migration from pre-RFC-6 to post-RFC-6 could require changes to stored array data rather than just metadata. It would be valuable to determine how much, if any, written data has made use of the multi-multiscales feature.
Comment authors#
This comment was written by: Chris Barnes, Davis Bennett.