Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: DM, pipe_tasks
-
Labels:None
-
Story Points:1
-
Epic Link:
-
Team:Data Release Production
-
Urgent?:No
Description
Matthew Becker has pointed out that the docstring for the spherematch query_self is inconsistent with the name/behavior. This will be fixed in an upcoming version of spherematch=0.10 (which is not released yet). At the same time, a new method will be added (find_groups) which will yield the behavior that we use/expect in isolatedStarAssociation.py.
This ticket encompasses two items to prepare for this change.
First, a new build of rubinenv will be created to pin spherematch<0.10 which will not change the existing env (therefore it is a new build and not a new version).
Second, updates will be made to isolatedStarAssociation.py that will be able to handle both versions of spherematch.
A follow-up ticket will be able to unpin spherematch in rubinenv at our leisure.
Attachments
Activity
Just to clarify, the functionality that is called query_self in v0.9 has been moved to query_groups in upcoming v0.10. And this is the functionality that we need. So if query_groups is available that means we should use it (new version/new name), otherwise fall back to query_self (old version/old name). Our env is currently pinned to the old version so that when the new version comes out nothing will break. After it comes out, we can release the pin with the new version of rubinenv and everything should transition smoothly.
BTW, I did test locally that this code works with the new version of smatch.
Oh, I see. I looked at the commit that added query_groups and didn't see any changes to query_self, so I assumed that it had the same behavior and query_groups was improving the behavior. I didn't realize that it was part of a larger PR that contained the changes to query_self.
Fred Moolekamp
added a comment - Oh, I see. I looked at the commit that added query_groups and didn't see any changes to query_self , so I assumed that it had the same behavior and query_groups was improving the behavior. I didn't realize that it was part of a larger PR that contained the changes to query_self .
This looks fine. From the looks of it, the main change is that query_group removes duplicate entries, is that right? If so, do they have an ETA on the release of 0.10? I'm just wondering it's ok to leave the current behavior, or if should we pin https://github.com/esheldon/smatch/tree/fd01e0c679f5ca516c747679022d32f4d12086f4 to make sure that we are using the correct algorithm.