Closed spoorendonk closed 4 years ago
I updated the Travis setup a while ago, but this PR was built before doing that. I'm going to close and er-open to force a re-build.
Did a rebase on top of master. It failed on a test on windows
now failing on mv: rename BLAS/*.f to ./*.f: No such file or directory
for osx10?
The Windows failure seems to be a real failure, but I guess it can't really be due to this change. I think I may have seen that one before. It's basically getting the wrong optimal solution value. It could be a bug that was already there and it just getting triggered by a slightly different execution path once the assert is removed or perhaps some numerics are a tiny bit different. It's weird. I'll see if I can reproduce.
We may as well do this one, too.
The assert checks for initial variables and fails in debug mode if no variables are present. It is OK to have no variables - it is an unnecessary check. The master restores feasibility using artificial variables.
It stalled on the macOS 9.2 build. Timeout? I have no idea why it did not succeed.