jerch

Hey @olivierdalang Yes that can be done, but the way to declare the intention and resolve it correctly will get cumbersome. Will have to think about it for some time,...

Oh well again missing tests 😅 otherwise I'd have spotted that (e.g. updatedata has a few call_command test cases).

At a first glance - yes this looks like an issue with the mro. Can you re-check, whether for a successful run the mro looks the same? To further debug...

Eww thats a serious bug - yepp, the mro must be stable for a certain dependency configuration, it may never flip positions.

@wadewilliams Sorry I had no time yet to look at the issue and where the mixup happens. In theory this should not have happened, because of how it is coded:...

@stevelacey Thx for the idea, for follow-up I gonna comment on #157 instead.

@hayder-mustafa-aswar I tried to provoke the glitching MRO issue but was not able to find a faulty pattern. Can you give me a stripped down code example, that still shows...

@wadewilliams Thx for the diagrams and MRO outputs. From the looks of the diagram the changing MRO here is still valid, as the fields have no sub dependencies on each...

@hayder-mustafa-aswar I cannot help you w'o a small repro, sorry.

@ all Gonna close this bug, as the initial issue cannot be repro'ed and the MRO discussion did not reveal anything buggy. Since there were different inputs from different ppl...