-
-
Notifications
You must be signed in to change notification settings - Fork 399
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix heasarc tests #2842
Fix heasarc tests #2842
Conversation
c1b7851
to
df4e08a
Compare
Codecov Report
@@ Coverage Diff @@
## main #2842 +/- ##
==========================================
- Coverage 66.51% 66.49% -0.03%
==========================================
Files 235 235
Lines 18096 18081 -15
==========================================
- Hits 12037 12023 -14
+ Misses 6059 6058 -1 see 1 file with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
@@ -15,6 +15,11 @@ | |||
|
|||
@parametrization_local_save_remote | |||
class TestHeasarc: | |||
|
|||
@pytest.fixture(autouse=True) | |||
def _patch_get(self, patch_get): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hmm, I think the original issue is that the mock should not have a remote conditional in it at the first place, the idea with all the remote tests is that they are used the same way a user would (aka without the mock patch)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hmm, I think the original issue is that the mock should not have a remote conditional in it at the first place, the idea with all the remote tests is that they are used the same way a user would (aka without the mock patch)
I am working on extending the heasarc services, and when I added a new test class, it was using patch_get
because it had autouse=True
, so I moved the autouse=True
part to the classes that need it. The class I am creating does not need it. Is there a better way to do it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I changed this to WIP because I see errors in my local testing and I thought I am fixing them, but they don't show up in the CI.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
re autouse: yes, removing it sounds good, most modules don't have autouse.
CI: we don't run the remote tests in CI on PRs, only once a week from cron, so that's why you may not see them failing. E.g. here I see the same heasarc failure as I see locally: https://github.com/astropy/astroquery/actions/runs/6279884195/job/17056292621#step:5:4890
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That being said, any improvements to the module is more than welcome, even to the point of a total gut out and starting from new (e.g. that's what I'm doing with the IRSA one, switching it altogether to VO backends). For heasarc, there are a few stalled PRs, too, but we can close them off if needed.
So, feel free to ping me on slack if there are any questions or if I can help with any of this.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
re autouse: yes, removing it sounds good, most modules don't have autouse.
CI: we don't run the remote tests in CI on PRs, only once a week from cron, so that's why you may not see them failing. E.g. here I see the same heasarc failure as I see locally: https://github.com/astropy/astroquery/actions/runs/6279884195/job/17056292621#step:5:4890
I see. That makes some sense. From that link I see both local and remote tests fail. Does test_mission_cols[local]
mean it is not using remote service?
../../.tox/py311-test-alldeps-devdeps-online/lib/python3.11/site-packages/astroquery/heasarc/tests/test_heasarc_remote_isdc.py::TestHeasarcISDC::test_mission_cols[local] FAILED [ 23%]
../../.tox/py311-test-alldeps-devdeps-online/lib/python3.11/site-packages/astroquery/heasarc/tests/test_heasarc_remote_isdc.py::TestHeasarcISDC::test_mission_cols[save] SKIPPED [ 23%]
../../.tox/py311-test-alldeps-devdeps-online/lib/python3.11/site-packages/astroquery/heasarc/tests/test_heasarc_remote_isdc.py::TestHeasarcISDC::test_mission_cols[remote] RERUN [ 23%]
../../.tox/py311-test-alldeps-devdeps-online/lib/python3.11/site-packages/astroquery/heasarc/tests/test_heasarc_remote_isdc.py::TestHeasarcISDC::test_mission_cols[remote] FAILED [ 23%]
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes, I think I overgeneralized above, in all the other modules, mocked and other local tests are separated from the remote ones, here it seems that everything has a local and a remote version. According to that, test_mission_cols[local]
should not use remote service.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is it obvious to you why test_mission_cols[local]
is failing in that link, but not in the CI test for the this PR?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
guesswork: I would think the canned version of that test integral query is outdated. And there is at least one bug (I guess more), with how the tests are run. E.g. a previous test method downloads the current version into the temp data dir (at least it doesn't go into the central cache location, that has been fixed already), and later tests pick that up instead of the canned version, even for the mocked version of the test.
Historically, I think this is one reason why we kept local and remote tests separate, less headache with what the given instances see, etc. (and we also don't need to fully duplicate everything local to remote).
E.g., if I remove all that uses integral_rev3_scw
, it's not failing any more:
astroquery/heasarc/tests/test_heasarc_remote_isdc.py .s..s..s..s..sF.s..s..s. [100%]
================================================================= FAILURES ==================================================================
_________________________________________________ TestHeasarcISDC.test_mission_cols[remote] _________________________________________________
self = <astroquery.heasarc.tests.test_heasarc_remote_isdc.TestHeasarcISDC object at 0x129c28670>
def test_mission_cols(self):
heasarc = Heasarc()
mission = 'integral_rev3_scw'
with self.isdc_context:
cols = heasarc.query_mission_cols(mission=mission)
assert len(cols) == 35
# Test that the cols list contains known names
assert 'SCW_ID' in cols
assert 'GOOD_ISGRI' in cols
assert 'RA_X' in cols
assert 'DEC_X' in cols
> assert '_SEARCH_OFFSET' in cols
E AssertionError: assert '_SEARCH_OFFSET' in ['SCW_ID', 'SCW_VER', 'SCW_TYPE', 'RA_X', 'DEC_X', 'OBS_ID', ...]
astroquery/heasarc/tests/test_heasarc_remote_isdc.py:133: AssertionError
========================================================== short test summary info ==========================================================
FAILED astroquery/heasarc/tests/test_heasarc_remote_isdc.py::TestHeasarcISDC::test_mission_cols[remote] - AssertionError: assert '_SEARCH_OFFSET' in ['SCW_ID', 'SCW_VER', 'SCW_TYPE', 'RA_X', 'DEC_X', 'OBS_ID', ...]
============================================ 1 failed, 33 passed, 17 skipped in 61.12s (0:01:01) ============================================
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is something I've been trying to work around also when experimenting with trying to switch to VOTable like in #2353. I found a solution that works 90% of the way to converting to VOTable but I ran into similar issues with the changes I made particularly for integral_rev3_scw
testing. Maybe I'll just submit a PR for it to get some potential input, but have been at it in some free time for a few months now and haven't gotten anywhere.
Not sure if this is helpful or not, but just thought that since there was a mention on extending heasarc I would share a quick note on my experiences.
@bsipocz, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The column name back-and-forth needs to be fixed by updating the name to pass the remote queries and thus the canned data file in the mock tests also needs the column name fixed.
fixture fixes look good, and I agree that reorganizing tests can be done in a follow-up.
assert 'SEARCH_OFFSET_' in cols | ||
# assert 'SEARCH_OFFSET_' in cols |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There has been a bit of a back-and-forth with this column name. Could you consolidate those changes to actually fix rather than skip the test? E.g. have a pass with the remote-data tests, and update the canned data file accordingly to get the mocked tests pass.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't know how to update the data files. I tried deleting them and running remote tests, but they didn't update
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
you need to either manually edit, or manually overwrite them with a freshly downloaded version (as ideally the datafiles we have canned up in the test suite are actually relevant versions of the remote responses).
df54874
to
8533c7d
Compare
3e98967
to
c4f6ff4
Compare
The remaining failure is from the cadc tests. |
Thank you so much @zoghbi-a! |
This a fix for the heasarc tests. Specifically:
patch_get
to class level to allow for other classes (in prep) that do not need it.