-
Notifications
You must be signed in to change notification settings - Fork 68
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fixes failing DV3D tests and VCS behavior #2003
Conversation
@doutriaux1 @aashish24 @danlipsa Please review. |
@chaosphere2112 Looks reasonable. I am OK with merging this if all but the diags tests pass. |
@doutriaux1 @aashish24 Ping. Can we merge this? I don't like to have failing tests on the master branch. |
@chaosphere2112 it fails for me after I merged master in. @danlipsa please try to merge master in and run from scratch. If it works for you then go ahead and merge. Thanks. |
@chaosphere2112 @doutriaux1 Indeed dv3d_ still fail for me as well. @chaosphere2112 any suggestions? |
@danlipsa @doutriaux1 What are the test failures you're getting? Sorry, been out. |
@chaosphere2112 This is the file generated on my machine for dv3d_vector_test. Don't you get the same results? |
@danlipsa Nope, it passes for me. Are you building from scratch, or just updating the |
@chaosphere2112 remember it didn't work for me either running from scratch. I'll merge master in from here, maybe that will help. |
@doutriaux1 I'll try and build from scratch to see if it stops working for me. |
@chaosphere2112 Thanks, updating DV3D made it pass for me. So, I am fine with merging this 👍 |
@danlipsa @chaosphere2112 if it works for both of you go ahead and merge. |
One small change in the DV3D package (changes how inherited parameters are initialized), some re-jiggering of logic in
vcs/dv3d.py
, and a small tweak in the testing framework.axes
be inherited from the parent as well (for children of Hovmoller3D).The commit (9404c34) that broke the integration did so because we can now instantiate a new DV3D graphics method using the
vcs.creategraphicsmethod
function (which seems reasonable to me).The way that this broke everything was that previously, that function returned
None
, which we later compared against in thecanvas.__plot
function and decided to just use the original graphics method (presumably a fallback of some sort). This made the behavior of the 3D plots emulate the normal DV3D behavior (where multiple cells are linked together), but violate the standard VCS behavior (where everything is separate).We can argue about that behavior separately; this pull request does not affect it at all, and I think that the offending commit is also fine (because the functionality it adds is good). If we want the old behavior, we should rewrite the logic in
__plot
so the change is in the relevant place, rather than crippling other functionality because it happens to tweak the behavior.