Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.48%. Comparing base (6f61dc3) to head (173acaf).
⚠️ Report is 14 commits behind head on master.

@@            Coverage Diff             @@
##           master     #397      +/-   ##
==========================================
- Coverage   99.91%   99.48%   -0.44%     
==========================================
  Files           8        8              
  Lines        1147     1154       +7     
==========================================
+ Hits         1146     1148       +2     
- Misses          1        6       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

I think it would be a good opportunity to also clean up existing VERSION checks in the package and the tests.

It actually seems that Julia < 1.10.10 is completely untested at this point:

I think testing 1.10.0 in addition to 1.10.10 is excessive. If a bug is ever triggered only in 1.10.0 it's not a good use of anyones time to fix it.

If a bug is ever triggered only in 1.10.0 it's not a good use of anyones time to fix it.

In principle, you don't have to fix anything apart from correcting the Julia compat entry: If the package does not support 1.10.0, it should not be declared to be compatible with 1.10.0.

Currently, the problem is even more sever as not testing min means that not only versions 1.10.0 - 1.10.9 but all of 1.6.0 - 1.10.9 are not tested. If you'd like to reduce CI times, IMO it would actually be better to skip pre and/or to test min and 1 instead of lts and 1 (based on the assumption that if e.g. tests pass with 1.10.0 they most likely also pass with 1.10.10).

I agree that in theory we should be testing v1.6.x right now. But one can argue if a version of Julia is no longer supported then there shouldn't be an expectation that packages continue to work

I'm less worried about CI time than I am additional developer burden. Every extra test is an extra point of failure...

can this be closed?