Thomas Robitaille
Thomas Robitaille
This fix looks good but I can investigate the test failure tomorrow. Looks puzzling at first glance!
The failure is related, it seems to be related to the fact if e.g. crval is modified in place then flag doesn't get reset. I need to look more closely...
I think this specific example changes one element of the array in-place so I'm not sure if the setter gets called?
Another option on the astropy side might be to use a numpy array subclass to store the wcsprm elements so that we can correctly set the flag if individual items...
I think https://github.com/astropy/astropy/pull/16411 might be the correct fix now - that PR is still waiting for a bugfix WCSLIB release (have pinged Mark about it) but once that's done I...
@WilliamJamieson - just to check, does it fail for you if you check out the #16307 merge commit? There was another PR after that made some tweaks, though I have...
Not sure if this is related but I noticed that the configuration value is accessed in the ``__init__`` call which means it is evaluated at import time. Is there are...
Just to put it on the table, an alternative is to compute a checksum of the wcsprm, but that would be more work as there are a lot of attributes...
@manodeep - I've now rebased on top of the WCSLIB 8.3 PR, and have then made it so we check the checksum prior to calling wcsset. The last commit is,...
The WCSLIB bug fixes were approved by Mark - he said he will not release 8.3.1 yet but we can always hold off from merging this until it is if...