Different behavior of MIMO system construction on Julia 1.12

[franckgaga]

Hello!

I don't know if it's intended (probably not) but on Julia v1.12, this code:

sys = [tf(1) tf(2); tf(3) tf(4)]

produces:

2×2 Matrix{TransferFunction{Continuous, ControlSystemsBase.SisoRational{Int64}}}:

...

whereas on Julia v1.11 and below the result is:

TransferFunction{Continuous, ControlSystemsBase.SisoRational{Int64}}

...

This is also brings up #975, since a CI test run on the new pre channel would have probably catch this.

edit: just tested using Pkg; Pkg.test("ControlSystemsBase") and yes many tests no longer pass of v1.12

[baggepinnen]

Thanks for reporting the issue, I'll try to have a look at it tomorrow. We have historically tested on nightly as well, but nightly was so often broken that the test ended up being useless, so many false positives that we stopped looking at it. When no one looks at the failing tests due to them always failing for the wrong reasons, they're no longer useful.

[franckgaga]

I had the same experience with nightly, pre seems to be a better compromise with much less false positive.