Closed richlander closed 3 months ago
@michaelgsharp was already doing this in https://github.com/dotnet/machinelearning/pull/7034. I think we require more changes than just retargeting because of differences in test output. I'll defer to @michaelgsharp's call though.
I see, we were not changing the build machines in that other PR - only helix queues.
I think it should be just fine. There are some fixes I need to do to update the Windows queues, but this doesn't change those so that won't hit it. We also build our official bits for Linux on CentOS machines, so this shouldn't have any problems/issues in that area either.
I also filed this issue: https://github.com/dotnet/machinelearning/issues/7088
I commented on the other PR.
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 68.47%. Comparing base (
8b483f4
) to head (efec8e6
). Report is 3 commits behind head on main.
Looks like the issue is on the Windows leg so I'll let you folks look at that.
BuildAnalysis linked the failure to https://github.com/dotnet/machinelearning/issues/7067 which is a known issue in auto-ml tests. Merging.
We are deprecating
Ubuntu.1804.Amd64.Open
.