Closed corrado9999 closed 3 years ago
Another option that is good for me:
xfail
)Option 4 is what it currently implemented. The drawback is that, if you make some changes that make the test fail, it will pass by unnoticed. @aurghs also told me that the incoming changes of update-README.md
branch will make pytest --doctest
in GitHub actions fail if the burst data are not available, as there is an example in the README.md showing how to access a burst.
Thus, I am opening a branch to upload sample data files and remove the xfail
mark.
For testing burst opening, we need real data to pass to rioxarray. Of course using the original TIFF files is not feasible, as they can easily be very big, order of GiBs.
I played a bit to solve this issue and I have a few options:
.gitignore
to avoid mistakes and we could also "cache" the produced file to avoid generating it repeatedly.Personally I think the second option is not worth losing the original blocksize, while the third option is more code to maintain and it slightly complicates the tests. Nonetheless, the third option could be useful should we need more meainingful data to put in, for example for calibration.