Debug endless docs build #165
Conversation
| "body_widget.output_selection_box.typical_outputs_widget.value = [Histogram]\n", | ||
| "body_widget.parameter_box.parameter_refresh_button.click()\n", | ||
| "body_widget.result_box.run_button.click()" | ||
| "ess_widget.children[0].children[0].value = RandomDistributionWorkflow" |
There was a problem hiding this comment.
I think it's fine as a fix for now.
I don't want to spend time understanding the underlying issue with ipywidgets, threads, event loops and trailtets...
jl-wynen
force-pushed
the
debug-endless-docs-build
branch
from
64572ee to
12c6612
Compare
This is needed to prevent the docs build from freezing. At least in my tests.
jl-wynen
force-pushed
the
debug-endless-docs-build
branch
from
12c6612 to
cdec165
Compare
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "Note that the order of operations here is important.\n", |
There was a problem hiding this comment.
All of the code that interacted with the widget used to be hidden in the docs (with "nbsphinx": "hidden") because I didn't really want to have ugly things like ess_widget.children[0].children[0].value confuse the users.
Do we still want to hide stuff or do you think it's better to not hide for the sake of transparency?
There was a problem hiding this comment.
Hm, maybe that has something to do with the problem 🤔. Let me investigate.
Do we still want to hide stuff or do you think it's better to not hide for the sake of transparency?
These are developer docs. So I think showing the code makes sense in this particular instance.
There was a problem hiding this comment.
Hiding the cell has no impact on whether the build freezes. With the old code, it freezes regardless of whether the attribute is set.
It looks like switching the order of function calls around in the notebook avoids the freeze. But I don't understand why.
Note that the tmat setup needs to be removed before merging.