Julian Hall
So, I've reproduced the behaviour for [highspy_speed_test.py](https://gist.github.com/aphi/4b2f7cfc7ccc52465bead3c24ee9c3f4#file-highspy_speed_test-py) and now see what "converting to a Python list" means. Maybe this is a known `pybind11` issue, for which there is a solution...
Anyone have an idea how to fix it upstream? I'd have thought it was a known pybind11 issue
I've submitted a query to pybind11
Only solution at the moment is a warning in the documentation
As @odow says, the order of the rows shouldn't matter but, in general, it leads to solvers behaving differently. If the problem has a non-unique solution, interchanging the rows can...
Thanks, I'd rather fix `master` directly than have to learn from `v1.2.2`. However, this looks like a different issue
Assertion `infeasible_ == 0' failing is not an error, merely an indication that infeasibility could have been detected earlier, so just a minor inefficiency. In release mode, MIP should solve...
It's an assert, so "identifiable" termination, rather than something like a segfault and, by default, HiGHS builds in release. If folk build in `RelWithDebug` then they know that this can...
Fair point: I'll add a comment to the code for now to indicate that this is an efficiency warning rather than an error, and I guess it's only fair to...
Thanks. Overall, this is another very useful contribution to HiGHS from you. I will look at this in a week's time. I'm hard pressed with work this week