TST: Tests failures in coordinates caused by IndexError in IERS with numpy-dev #9240
Comments
|
I restarted https://travis-ci.org/astropy/astropy/jobs/585668145 to see if it was transient server problem. I cannot reproduce the failures locally. |
|
It repeated so does look like a real bug. Strange that it only turns up in doctests! |
|
I also cannot seem to reproduce the failures (tried both latest numpy-dev (1.18.0.dev0+6e75b9a) and the one used by travis (1.18.0.dev0+07a7eb5)... |
|
Hang on... I am going to try clear Travis CI cache and rerun it again... 🤞 |
|
Looks like some sort of corrupted cache. It is green now. Sorry for the noise! |
|
Fortunately, I had started on something else... But while testing that, I found a similar issues, which resulted from the IERS-A table not downloading completely. I think we do need to circumvent that by having a cached version or so... |
|
@mhvk , Travis CI is already doing some caching. But if you are talking about the other IERS issues, then I don't have anything useful to contribute. LoL... |
|
Ah, that means it might have been the same problem: for me, it seems the |
|
Yes, I do suspect that Travis CI cached a corrupted IERS table. |
A blame on the LOC pointed to #9226 though the PR itself was green. 🤷♀️
@mhvk , any idea?
Example log: https://travis-ci.org/astropy/astropy/jobs/585668145
UPDATE
Looks like numpy-dev job did pass in https://travis-ci.org/astropy/astropy/builds/585637569 ("about 2 hours ago" at the time of writing this with
numpy 1.18.0.dev0+07a7eb5) but failed in https://travis-ci.org/astropy/astropy/jobs/585641053 (also "about 2 hours ago" with1.18.0.dev0+07a7eb5). The PR merged seems unrelated to IERS nor coordinates. And no changes were merged into numpy-dev since 8 hours ago. Then, more tests failed on our end in the next run (https://travis-ci.org/astropy/astropy/jobs/585668145). Did IERS data server change something?The text was updated successfully, but these errors were encountered: