DM-54976: move obs_info to VisitImage and unify formatters further#41
Conversation
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## main #41 +/- ##
==========================================
- Coverage 75.12% 74.09% -1.04%
==========================================
Files 92 89 -3
Lines 10904 10514 -390
==========================================
- Hits 8192 7790 -402
- Misses 2712 2724 +12 ☔ View full report in Codecov by Sentry. |
timj
left a comment
timj
left a comment
There was a problem hiding this comment.
Looks good. So the formatter tests are now all implicitly in the RoundTrip code?
| yield archive, tree | ||
| case ".json": | ||
| tree_type = pytype._get_archive_tree_type(JsonRef) | ||
| tree = tree_type.model_validate_json(ResourcePath(uri).read()) |
There was a problem hiding this comment.
I wonder how well this works with the 600MB JSON visit image I made...
There was a problem hiding this comment.
"inefficiently", especially in terms of memory consumption, I imagine.
TallJimbo
force-pushed
the
tickets/DM-54910
branch
2 times, most recently
from
b46d392 to
acb1d81
Compare
Yes. I can't claim we're still touching all of the combinations of things we were before, and we probably want to beef up that kind of coverage again in the future. But I think the coverage is still pretty good right now via the RoundTrip stuff. |
TallJimbo
force-pushed
the
tickets/DM-54910
branch
2 times, most recently
from
599e392 to
dd3cf98
Compare
This requires a small change to every ArchiveTree implementation, which makes it a very broad change. Some LLM-generated tests have been deleted because they were focused on private interfaces that have gone away in this change. I think the high-level behavior is still well covered, if not necessarily the outer product of serializable type vs. archive format. Since the any-format, specific-serialized-type formatters haven't been around long, I've deleted them and made the old format-specific formatter stubs redirect straight to the new one.
The only butler datasets with the old formatters should be ExtendedPsf things (which are experimental), and we've agreed we can just go delete those.
TallJimbo
force-pushed
the
tickets/DM-54976
branch
from
e51e815 to
a0873e9
Compare
This could break depending on the order in which tests were run.
TallJimbo
force-pushed
the
tickets/DM-54976
branch
from
7be1616 to
698f380
Compare
2 participants
Requires lsst/daf_butler#1375
Checklist
doc/changes