You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Dec 1, 2022. It is now read-only.
In the discussion about Tiling (#29) we started talking about the problem of how to handle large coverages, raising the question of how we could provide the client with different views of the coverage at different resolutions, for greater efficiency and scalability. There are several possible solutions:
Link the coverage to an API that performs subsetting on the fly
Provide multiple versions of the range within the coverage (e.g. one version for each resolution or zoom level).
Create a special kind of Coverage Collection that collects together different versions of (what is notionally) the same coverage.
Option 1 is of course possible (and is already done in the OPeNDAP world) but I feel that we should have an alternative that doesn't require running a dynamic server. I think we rejected option 2 in the discussion of #29 because it can break the domain-range contract. Option 3 is probably the preferred option. This ticket is to develop that proposal.
I think the mechanism @letmaik proposed in #29 is OK, and I'll reproduce it here:
Should there be some metadata to indicate the type of collection we have here (i.e. multiple versions of the same coverage)? Maybe the type could be called a "pyramid" or something, borrowing the term from "image pyramid".
Could we also allow the full-resolution coverage to be published on its own, with a link back to this Collection? In that way a client could start by loading the coverage metadata, then find out automatically whether a "pyramid" option is available if it decides it can't load the whole coverage. Something like this:
This seems quite useful, though my guess is that in most environments it is actually the server doing the subsetting and CovJSON is just a mechanism to get the data across. In any case, this is easily doable in a future version in a backwards-compatible way, so I'm going to close this and wait until someone has a concrete need for it that can drive the discussion.
In the discussion about Tiling (#29) we started talking about the problem of how to handle large coverages, raising the question of how we could provide the client with different views of the coverage at different resolutions, for greater efficiency and scalability. There are several possible solutions:
Option 1 is of course possible (and is already done in the OPeNDAP world) but I feel that we should have an alternative that doesn't require running a dynamic server. I think we rejected option 2 in the discussion of #29 because it can break the domain-range contract. Option 3 is probably the preferred option. This ticket is to develop that proposal.
I think the mechanism @letmaik proposed in #29 is OK, and I'll reproduce it here:
The text was updated successfully, but these errors were encountered: