Misc fix #43
Misc fix #43
Conversation
jourdain
requested review from
alexcjohnson,
Marc-Andre-Rivet and
xhluca
as code owners
| return None | ||
|
|
||
| # Extract mesh | ||
| points = b64_encode_numpy(vtk_to_numpy(polydata.GetPoints().GetData())) |
There was a problem hiding this comment.
It will be either float32 or float64.
| # We have a numpy array that is compatible with JavaScript typed | ||
| # arrays | ||
| buffer = base64.b64encode(memoryview(obj.ravel(order="C"))).decode("utf-8") | ||
| return {"bvals": buffer, "dtype": str(dtype), "shape": obj.shape} |
There was a problem hiding this comment.
My understanding is that the original code was meant for 1D numpy array. I don't think that creates any problem for obj.shape (since a python tuple should be easy to convert to JS), but i'm curious whether there will be needs to do any modification on the JS side for that to work...
There was a problem hiding this comment.
On the vtk side we expect 1D, we don't care about the shape... Just kept it for @nicolaskruchten . ;-)
|
@jourdain The PR is merged; I think what you added looks good on the python side, do you have anything else in mind on the JS side? |
|
If we use [email protected] we are good... |
|
@xhlulu I let you merge it and eventually create a version 1.0.0 is everything is working as expected. Normally the F1 demo should be faster with that one. |
|
@jourdain Do you get this error locally?
|
|
I guess I'll have to try that specific example... |
|
Yeah it's a regression:
|
|
It looks like the cell connectivity get messed up. Along with some point coordinates not being correct... |
|
@xhlulu that should be good now... Good catch! |
|
@jourdain awesome. Updating the change log and will create a release afterwards. |
Need to rebase from the async but fix css typo in demos and add support for encoded binary arrays.
closes #42