Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: ip_isr
-
Labels:
-
Story Points:0.5
-
Epic Link:
-
Sprint:Alert Production S17 - 2
-
Team:Alert Production
Description
E.g. when running ip_isr's examples/runAssembleTask.py:
Traceback (most recent call last):
|
File "runAssembleTask.py", line 36, in <module>
|
runAssembler()
|
File "runAssembleTask.py", line 21, in runAssembler
|
assembledExposure = assembleTask.assembleCcd(assemblyInput)
|
File "/scratch/swinbank/ip_isr/python/lsst/ip/isr/assembleCcdTask.py", line 200, in assembleCcd
|
ccd = next(assembleInput.values()).getDetector()
|
TypeError: list object is not an iterator
|
Looks like a bad conversion to Python 3: it assumes that dict.values() returns an iterator, but in Python 2 it returns a list. Simple (but ugly) fix:
--- a/python/lsst/ip/isr/assembleCcdTask.py
|
+++ b/python/lsst/ip/isr/assembleCcdTask.py
|
@@ -197,7 +197,7 @@ class AssembleCcdTask(pipeBase.Task):
|
ccd = None
|
if hasattr(assembleInput, "has_key"):
|
# Get a detector object for this set of amps
|
- ccd = next(assembleInput.values()).getDetector()
|
+ ccd = next(iter(assembleInput.values())).getDetector()
|
# Sent a dictionary of input exposures, assume one amp per key keyed on amp name
|
|
|
def getNextExposure(amp):
|
Better suggestions welcome.
Rather shocking that this wasn't caught by tests. Sufficiently shocking that I'm wondering if I missed something. I suggest that rather than just committing the above tweak, this ticket should include a test which demonstrates that the task actually runs. An easy way to do that might be to simply migrate a slightly modified version of the code from examples/ to tests/.
Attachments
Issue Links
Activity
Sorry that I missed this till now. I bet we can find someone to do it around here.
Edit: Russell Owen said he'd take it on ASAP. We have a workshop tomorrow, but after that.
I have fixed on a ticket branch in DM-9370, but I don't know if this is the fix people want. It works, but it doesn't get around the fact that this should have been caught by tests. Not sure whether to submit for review or now...
I fixed the code but posponed a unit test for DM-9374. The fix is straightforward. Please see if this works for you.
Looks good to me!
PS This ticket is duplicated now by https://jira.lsstcorp.org/browse/DM-9370 which I just filed. Not sure whether to delete that or what, but presumable John Swinbank can help me clean up the mess :/