Karl Gordon
Karl Gordon
Or we could just store 500 x # of subgrid lnp values in the merged files. Shouldn't be too much space and this way we can put off the decision....
I think the procedure would be to have two conda environments, one for v1.4 and one for the dev version - v2.0. Does this make sense? This was the idea...
Apologies, I hadn't pushed the last few commits I did to the v1.4 branch for the v1.4 release. Doh! If you pull the v1.4 branch again, it should now report...
Not sure. This sounds like it will do many of the same tests as are already done. Unless there is a way to just test the parts that are not...
Go for it. We can always disable tests that take too long. :-)
You can see what is being used by looking at the testing code for this step. Specifically https://github.com/BEAST-Fitting/beast/blob/master/beast/physicsmodel/stars/tests/test_spectral_grid.py
Other test code might also need to be consulted. Check them in out in the beast/physicsmodel/tests and beast/physicsmodel/stars/tests.
This sounds like a fine plan. Easiest if you do this in a separate PR just for this change. Then it should be quick to get merged.
Yes. See #395.