DM-31924: implement new data ID packer and deprecate ExposureIdInfo #450
Conversation
|
A general comment for consideration, the |
|
That's a very good point. I put it in obs_base because I was thinking "this is a hook that obs_ packages may override", but I don't think it needs afw or anything else not available in pipe_base. |
TallJimbo
force-pushed
the
tickets/DM-31924
branch
from
6158e64
to
7c54d73
Compare
|
Yes, I moved the core Instrument functionality to pipe_base because it seemed like we really needed some form of |
TallJimbo
force-pushed
the
tickets/DM-31924
branch
from
7c54d73
to
5949dd5
Compare
|
Ok, I've moved everything on this ticket but the deprecation of old code to pipe_base. |
TallJimbo
force-pushed
the
tickets/DM-31924
branch
from
5949dd5
to
aba56d6
Compare
The replacement, CatalogIdPacker, is very similar, but it also has the ability to unpack Source/Object IDs fully, and that's not something we could add to ExposureIdInfo without breaking backwards compatibility in its constructor signature. Moving to a new but very similar class has two more smaller advantages: - it makes the transition from registry-provided dimension packers to configurable dimension packers easier by fully swapping out the helper classes that make the dimension packers; - it eliminates one case of too-overloaded "Exposure" nomenclature. I've moved the replacement to meas_base to avoid introducing a dependency (in either direction) between obs_base and skymap; I don't think those dependencies would be a big problem, but it's nice to avoid dependencies when you can, and meas_base already fits the requirements for where we'd put the new class: - It already depends on both afw and daf_butler. - It already depends on skymap and obs_base (either order for these relationships would acceptable, as we can register configurables with a pex_config registry either when we define the registry or when we define a particular configurable). - It is already a dependency of all Tasks that would use it (some of which live in meas_base).
TallJimbo
force-pushed
the
tickets/DM-31924
branch
2 times, most recently
from
5aee9e0
to
a184f7d
Compare
Checklist
doc/changes