Strip the IPY_MODEL_ prefix from widget IDs before referencing them. #6377
Conversation
|
@jdfreder, @ellisonbg, ready for review. I can't believe I made this silly mistake. |
|
I'm confused, why haven't the child serialization tests been failing in master? Edit: Would you be able to write a test for this? It seems like an important piece of the code to be untested for regressions. |
|
I'd love to write a test. Do you want to write a nose testing framework for the python side of widgets? If I had, say, a mock Comm object that I could plug into a widget, and then send/receive messages on this mock comm instance, it would be easy to test this. |
|
By the way, I'm a little confused why this wasn't noticed before too. Perhaps we haven't seen a case where js was sending back a reference to a model? |
As of now, I've been writing these in the CasperJS tests as embedded strings. Each time I need to test something in Python, I run the string as a cell, use Python print, and then use JS to check the output of that cell. It's ugly, painful, and brittle, but it works. 🐴 |
Yeah, I guess that makes sense. We must not be echoing back the values after they're sent to the front-end, which is a good thing and is as designed. 😄 |
|
(just want to make sure this doesn't fall off the table). I'm happy to write a test once we have a framework to test the python side of widgets (I think basically all we need is a mock Comm object). Writing such a framework is on my TODO list, but not near the top :(. |
|
Sorry, I was on vacation the past two weeks, I really should have merged this before I had left. I'll open an issue to remind you and/or I to write a test for this (low priority). |
Strip the IPY_MODEL_ prefix from widget IDs before referencing them.
|
See #6485 for tests |
Strip the IPY_MODEL_ prefix from widget IDs before referencing them.
This was a silly error, probably due to me.