Details
-
Type:
Story
-
Status: Invalid
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: daf_butler
-
Labels:
-
Story Points:8
-
Epic Link:
-
Sprint:BG3_S18_04, BG3_S18_05, BG3_F18_06, BG3_F18_07, BG3_F18_08, BG3_F18_09, BG3_F18_10, BG3_F18_11, BG3_S19_01
-
Team:Data Release Production
Description
Gen2 repositories already have SkyPix datasets (reference catalogs) that use possibly-different levels of HTM. That'll be tricky to support with our (probably naive) plan to have a single SkyPix pixelization and level in each Registry. Come up with a design that solves that problem, and at least think about whether HiPS/MOC datasets pose a similar problem (they probably do).
Options include:
- RFC rewriting reference catalogs to a common level.
- Have a common level for all join tables, but allow Datasets to use other levels (with either another join table or an in-database function to relate them).
We also need to think about how to deal with HTM vs. HEALPix; we need support for HTM for reference catalogs now (and already have the necessary low-level code in sphgeom), but we know we'll need to do HEALPix in the future. That will probably require adding HEALPix to sphgeom (backed by third-party HEALPix libraries, but still not trivial), which we'd rather not do now (not part of this ticket). But once HEALPix is available, do we do a big-bang switch to it and drop HTM, or do we design a system that permits both to exist within the same registry?
Outputs: write up a proposed design (Confluence or DMTN), and RFC it.
Attachments
Issue Links
- is contained by
-
DM-17023 Refactor the Dimensions and query system
- Done
As part of this ticket, we should define some named constants to avoid special "SkyPix" or "skypix" string literals in the Registry classes.