Closed sbailey closed 5 years ago
See also desihub/desisim#391. There is no reason not to have a fallback data file cached to avoid this.
We already use a cached data file by default (via freeze_iers) to avoid problems like this and enforce a consistent model of future polar motions for coordinate transformations.
The failing unit test is exercising the code to occasionally update the cached data file, so it is useful to know that it is failing but this should have no impact on code that uses freeze_iers
.
There is already an astropy issue for this: astropy/astropy#8304.
Ah, fair enough. While we have people's attention though, let me note that the desisim issue I referenced is still open. In other words, desisurvey is using freeze_iers
, but every DESI package using it?
@weaverba137 other packages are not yet using freeze_iers
, but it hasn't been problematic enough to become a priority compared to other development issues.
The navy.mil site is back and tests are working again so I'll close this ticket. If this becomes problematic again in the future, see astropy/astropy#8308 for how they used the mirror site http://toshi.nofs.navy.mil/ser7/finals2000A.all
unit tests are failing on the IERS download, since http://maia.usno.navy.mil/ser7/finals2000A.all is offline. Check whether this has moved somewhere else or whether we just need to wait for it to come back. I get slightly different errors on my laptop vs. NERSC, but one version is: