search

NOAA-OWP / wres

Code and scripts for the Water Resources Evaluation Service
Other
2 stars 1 forks source link

Scenario 510 has failing .svg comparison #360

Open epag opened 5 hours ago

epag commented 5 hours ago

Apparently the scenario 510 test passes or fails depending on the hardware that runs it. We believe this is related to JFreeChart FAKE2_FAKE2_MEAN_ERROR FAKE2_FAKE2_MEAN_ERROR_created FAKE2_FAKE2_SAMPLE_SIZE FAKE2_FAKE2_SAMPLE_SIZE_created ![Uploading Screenshot 2023-04-26 155714.png…]()

HankHerr-NOAA commented 4 hours ago

FYI... I tried to reproduce the problem on the RHEL 8 owpal-d-ised01 machine, but it matched the benchmarks. Evan described how I can use Jenkins to reproduce it for investigation purposes.

Given the nature of the differences - tick marks on the domain axis differing - only one of those plots will actually match the data. That's presumably the benchmark. We need to confirm that. We also need to see if the PNG looks correct (i.e., does it have the same incorrect tick marks as the SVG?). Finally, what is production producing?

I'll take a brief look on Tuesday or Wednesday of next week. Have a great weekend!

Hank

james-d-brown commented 3 hours ago

Interesting, well that provides a hook to dig into w/r to how tick marks are autogenerated, but it must be a platform dependent "feature". As an aside, we are miles behind w/ jfreesvg versions because later versions are not deterministic w/r to the string produced, but that seems to be a separate issue from the one reported here, which is probably not related to jfreesvg, rather jfreechart.