You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently the RecSec tests just plot code functionality but have no check on if the figures look how we want. Some simple checks are to run record section for one of the example events and make sure the data look okay. This is prone to failure as there are a lot of options and option combinations (and growing by the second). By introducing both data and synthetics, mismatching waveform lengths, time shifts etc., it is very easy to slip into a realm where code is executing but figures do not look right.
We need a set of control data (data and synthetics) and a test suite that generates figures for a variety of parameter options/combinations that we might experience in the real world. We can then use the Matplotlib test kit to check resulting .png files against known test results.
Parameter combinations that are testworthy (add to this as we see fit):
Data and synthetic comparisons with a move out and xlimits
Data and synthetic comparisons where trace start and end times (and sampling rate) are not the same
The text was updated successfully, but these errors were encountered:
bch0w
changed the title
recsec test data: a collection of figures with various options
recsec test data: visually checking a collection of figures created with various options
Oct 20, 2022
Currently the RecSec tests just plot code functionality but have no check on if the figures look how we want. Some simple checks are to run record section for one of the example events and make sure the data look okay. This is prone to failure as there are a lot of options and option combinations (and growing by the second). By introducing both data and synthetics, mismatching waveform lengths, time shifts etc., it is very easy to slip into a realm where code is executing but figures do not look right.
We need a set of control data (data and synthetics) and a test suite that generates figures for a variety of parameter options/combinations that we might experience in the real world. We can then use the Matplotlib test kit to check resulting .png files against known test results.
Parameter combinations that are testworthy (add to this as we see fit):
The text was updated successfully, but these errors were encountered: