Fix granule duration check, to allow shorter than 48 scans in a granule #23
Conversation
Signed-off-by: Adam.Dybbroe <a000680@c21856.ad.smhi.se>
Codecov Report
@@ Coverage Diff @@
## master #23 +/- ##
==========================================
+ Coverage 27.80% 29.71% +1.91%
==========================================
Files 11 11
Lines 1705 1740 +35
==========================================
+ Hits 474 517 +43
+ Misses 1231 1223 -8
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
Signed-off-by: Adam.Dybbroe <a000680@c21856.ad.smhi.se>
Signed-off-by: Adam.Dybbroe <a000680@c21856.ad.smhi.se>
|
@TAlonglong and @mraspaud Can we merge this one? |
|
Not yet, tests are failing and it hasn't been reviewed :) |
| posttroll_message = PostTrollMessage(0, metadata) | ||
| delta_t = timedelta(seconds=82.8) | ||
|
|
||
| with patch.object(PostTrollMessage, 'get_granule_duration', return_value=delta_t) as mock_method: | ||
| result = posttroll_message.is_segment() | ||
| self.assertTrue(result) |
There was a problem hiding this comment.
Could you also add a check for 48 scans?
There was a problem hiding this comment.
I thought I had!?
I have a test for 82.2 seconds (47 scans) and 84.1 sec (48 scans).
The thing is, as far as I interpret things, that in the SDR files you have the start and end time in terms of time of the first scan, and time of the last scan. But to get to the duration of the entire granule you need to add the time length of one scan (~1.8 sec). Thus, the duration of 48 scans is ~85.9 sec and ~84.1 seconds for 47 scans.
There was a problem hiding this comment.
I don't see the test for 84.1 seconds in test_is_segment, what am I missing?
There was a problem hiding this comment.
There was a problem hiding this comment.
@mraspaud Do you see it now?
The couple of test-suites that fail (ubuntu/wondows latest), doesn't have anything to do with this PR as I see it. |
How do you mean? Should we omit PPS processing all together if there are 46 scans or less? |
Could you just add a |
I'm just unsure what the format specification says about viirs granules. Is is just 47 or 48 scans, or can it be less or more? |
…culation Signed-off-by: Adam.Dybbroe <a000680@c21856.ad.smhi.se>
Yes, me too. From our own processing it looks like it is either 47 or 48 scans, but the actual length varies a bit more than just between two values. So, seems like 48 scans is not always exactly the same size in terms of start and end times in the files! I get betwee 1min24.1sec to 1min24.5 sec (that is actually just end-start times - so missing 1.779 sec) for the 48 scan files. I have linked the SDR spec already, but don't find the allowed/occuring number of granules. |
|
I have adjusted the granule length calculation adding length of one granule. |
Signed-off-by: Adam.Dybbroe <a000680@c21856.ad.smhi.se>
|
About skipping, I meant just skipping these specific tests, with https://docs.pytest.org/en/latest/skipping.html |
Okay, thanks, but now I don't see what exactly was failing anymore, but it was something about gdal, and it was windows and ubuntu experimental, so I rather want to fix it in a separate PR then. The github ci was added just recently. |
|
@mraspaud Where are we with this? |
Signed-off-by: Adam.Dybbroe a000680@c21856.ad.smhi.se
pytestflake8 nwcsafpps_runner