The version 2.1.0.6 is tagged but not actually released

[mhvk]

Weird. Found via discussion at and below astropy/astropy#18551 (comment).

Maybe good to create a new 2.0.1.7 ahead of astropy 8.0, with the current fixes.

[pllim]

Indeed looks like the publish workflow failed to upload because this manylinux aarch64 job failed. The log has since expired so details are unavailable.

https://github.com/liberfa/pyerfa/actions/runs/12998221846

Are you able to retrigger that job? If not, you would either have to manually run that release and upload yourself, or retag.

[neutrinoceros]

Out of curiosity, is there a reason for apparently freezing the first 3 components of the version number ? I would expect the next release to bump the second one, rather than the fourth.
I would also advise to let 2.0.1.6 be a non-release instead of reviving it a year later; even if you can retag or rerun the publishing job, both actions could have unexpected consequences for users or maintainers.

[mhvk]

Yes, there is a reason, in that the first three are the version of the underlying ERFA library, and the last one for the wrappers. Of course, where it is a good reason is another question, but easier to just stick with it.

[neutrinoceros]

Ah, that makes sense. But then there's so little information in the last element that it might as well be calendar.