Join GitHub today
GitHub is home to over 36 million developers working together to host and review code, manage projects, and build software together.Sign up
This is a PR that should make it easier to navigate to the cell that was in error when you run into an error. Specifically, it allows you to click on a link that jumps you right to the cell where the error occurred.
How does it do this?
It does nothing if there is no error.
I think I'd want to include tests for the new apis that this introduces before we released this, but I want to iterate a couple of times first on the exact form this should take.
But nonetheless the existing tests needed to be modified to pass as of this latest commit.
@@ Coverage Diff @@ ## master #244 +/- ## ========================================== + Coverage 80.69% 90.64% +9.94% ========================================== Files 13 14 +1 Lines 1233 1101 -132 ========================================== + Hits 995 998 +3 + Misses 238 103 -135
This is a neat idea. I like the functionality but I am a little hesitant to add all the extra cells. From a presentation aspect it looks mostly right (maybe hiding the cell ID output as well from display) but from a model pov we have many extra cells. Classic notebook would render this oddly today and the cell execution counts would (I believe?) all be even numbers with the markdown cells in-between.
Do you think maybe we could insert a hidden html output on each cell which includes the cell id in the same manner as this? That's eliminate the cell count / visualization issue but let you stlll jump to failure outputs?
I'll see if I can chat 1:1 with you to see if we can iterate on the approach and desired end-point offline
The user interface in commuter looks good now. I do still worry about the logic of doubling the number of cells, especially with not all interfaces respecting hidden fields yet.
Iterating on the cells for an output notebook now would be quite different than iterating on the input cells. Would it be possible for us to add the display output to the existing cells rather than make new cells? Each cell could have a hidden html output linking to the next cell as it's final output. Thoughts on taking that approach?