[13:32:09] I've now made a fast implementation for the intermediate function, so I tried switching it into Z27011, but it's returning a whole lot of voids for both upper and lower bounds. (re @u99of9: I think this is now working. There is an intermediate function that could be much faster, but the current messy composition seem...) [13:41:16] The Z27008 function returns void by default, but that should be only when one of the two arguments supplied is itself void. (re @u99of9: I've now made a fast implementation for the intermediate function, so I tried switching it into Z27011, but it's returning a who...) [13:42:45] or sometimes there are unresolved references again... (top right fail at Z25985) : https://tools-static.wmflabs.org/bridgebot/9bf44fd9/file_72412.jpg [13:49:28] The tests at Z27011 don't seem resolve as normal. I see this after refreshing after reloading and achieving complete (failing) tests. : https://tools-static.wmflabs.org/bridgebot/03cf1829/file_72413.jpg [13:51:25] I guess this is going to trigger issues for Cory. Have fun, 🎃! [13:52:11] And me! 😎👍 (re @u99of9: I guess this is going to trigger issues for Cory. Have fun, 🎃!) [14:41:15] Hmm… I appear to have fixed it (all tests pass before Publish) … just trying to see whether I can work out how 😏 (re @u99of9: I guess this is going to trigger issues for Cory. Have fun, 🎃!) [18:05:47] So I think the root cause was Z20584. I gave that a new implementation with a Z801 wrapper and disconnected the other composition. Your version of Z27011 now fails only with mismatched results. The current version doesn’t have those mismatches, so I’ve left that in place for now. I can’t think why the function would give different results in the parts that haven’t changed [18:05:47] [18:05:47] , but there are stale results in the testing cache. All tests pass in testing for the current version when fresh results are provoked, so 🤷‍♂️ (re @Al: Hmm… I appear to have fixed it (all tests pass before Publish) … just trying to see whether I can work out how 😏) [18:44:52] Tests for Z25867 now fail (as does the function). Error given is Z503 but all the functions in the composition do have connected implementations. If True or False replaces the call to Z13740, the relevant cases pass, but direct calls to Z13740 fail. Python implementation tests report: [18:44:53] Perform test error: 'Z14102' isn't a known Object [18:44:55] (It was deleted in January) but, yeah, it’s still referenced in the function’s Z8K3. (re @Al: Hmm… I appear to have fixed it (all tests pass before Publish) … just trying to see whether I can work out how 😏) [18:52:00] (Implementations also list five test cases but only four are listed on the function page.) (re @Al: Tests for Z25867 now fail (as does the function). Error given is Z503 but all the functions in the composition do have connected...) [23:40:25] For me Z25867 passes all tests. But Z13740 is now having real trouble. Likely because of the ghost implementation. I'll remove that manually now to see if it improves things. : https://tools-static.wmflabs.org/bridgebot/a04aea2c/file_72435.jpg [23:46:02] Hmm, arguably even worse! : https://tools-static.wmflabs.org/bridgebot/1ef70c82/file_72436.jpg