Details
-
Type:
Story
-
Status: Invalid
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: pipe_drivers
-
Labels:None
-
Story Points:6
-
Epic Link:
-
Team:Data Release Production
Description
pipe_drivers currently does not have a test suite (for a number of reasons), but it needs one to prevent breakage. We could take some inspiration from the tests in pipe_tasks, and put something together in pipe_drivers. An alternative approach would be to modify ci_hsc to use the pipe_drivers scripts; that might allow us to break it out the scons mold that seems to annoy some people.
Attachments
Issue Links
1.
|
Minimal import and flake8 unit test for pipe_drivers |
|
Invalid | Unassigned |
2.
|
Create data- and ingest-only version of ci_hsc |
|
Invalid | Jim Bosch |
3.
|
Move CmdLineTask ci_hsc execution to a new package |
|
Invalid | Dan Taranu |
4.
|
Create pipe_drivers-based ci_hsc execution package |
|
Invalid | Yusra AlSayyad |
5.
|
Create new Gen3 conversion package for ci_hsc dataset |
|
Invalid | Jim Bosch |
Activity
Yusra AlSayyad, Sophie Reed, Dan Taranu: I've created some sub-task tickets to track the work we started last Thursday. Let's use DM-9059 as the integration branch for these, and merge that to master when all of the sub-tasks are done (except maybe DM-16263, which is pretty independent).
I think I've got the first piece (base ci_hsc data package that installs) done. Dan Taranu and Yusra AlSayyad, could you please try (on lsst-dev):
# <load LSST environment as usual>
|
export EUPS_PATH=/home/jbosch/local/20181012:$EUPS_PATH |
setup -v ci_hsc w.2018.42-3-gf5cd74e |
# <setup your derived ci_hsc_X with -k>
|
# <run your derived ci_hsc_X
|
That ci_hsc setup should bring in a custom version of sconsUtils as well (I needed to patch that to get installation of symlinks working properly).
If you're especially brave, you can take a look at the last commit to ci_hsc tickets/DM-9059 to see if you can make similar changes to your SConstruct to get it to install as well.
Is it safe to mark everything here as invalid or won't fix due to DM-19817?
I believe this has been invalidated by RFC-599 and related tickets.
We should at least make a minimal test that imports the Tasks and enable flake8 and Travis to prevent code with undefined variable names getting merged that require later fixing (e.g.
DM-15534).airglow:pipe_drivers yusra$ flake8 --ignore=E133,E226,E228,N802,N803,N806 | grep name./python/lsst/pipe/drivers/__init__.py:2:1: F403 'from .version import *' used; unable to detect undefined names./python/lsst/pipe/drivers/multiBandDriver.py:419:21: F821 undefined name 'dataRef'./python/lsst/pipe/drivers/multiBandDriver.py:429:36: F821 undefined name 'dataRef'./python/lsst/pipe/drivers/skyCorrection.py:197:17: F821 undefined name 'maskedImage'