Closed nimmolo closed 2 months ago
Hm. On one of the CI tests (only push!), the obs actually seems to get touched by the image transfer script.
Progress: |========================================= FAIL ReportTest#test_darwin_observations (158.91s)
--- expected
+++ actual
@@ -1 +1 @@
-["1046914674", "https://mushroomobserver.org/1046914674", "HumanObservation", "2006-05-12 17:21:00 UTC", "MushroomObserver", nil, "Fungi", nil, "Kingdom", "Fungi", nil, nil, "Mary Newbie", "2006-05-11", "2006", "5", "11", "USA", "California", nil, "Burbank", "34.185", "-118.33", "148", "294", "Found in a strange place... & with śtrangè characters™"]
+["1046914674", "https://mushroomobserver.org/1046914674", "HumanObservation", "2024-05-02 03:31:01 UTC", "MushroomObserver", nil, "Fungi", nil, "Kingdom", "Fungi", nil, nil, "Mary Newbie", "2006-05-11", "2006", "5", "11", "USA", "California", nil, "Burbank", "34.185", "-118.33", "148", "294", "Found in a strange place... & with śtrangè characters™"]
Changing the test to expect the fixture's current updated_at time, rather than the original updated_at, clears the error.
Oh i get it. The obs has been updated.
We seem to have obs
matrix_box
es that have been cached with a blank image. My working hypothesis is that the image was not done transferring to the image server at the moment where the cache was first created.This PR changes the image transfer scripts to
touch
the obs when transfer is done, thereby invalidating the cache and forcing a new one. Hopefully that resolves the blank image issue.TODO in a later PR: redo the image scripts as a queued job.