Update custom-components import paths and tests #8666
Conversation
There was a problem hiding this comment.
Code changes LGTM, but I'm guessing whether or not we want to revive this behavior (given it was an unintentional implication of how the Python module system works) ends up being on the product team to make a final decision.
If not many people are running into this, I'd lean toward not doing so given this was never supposed to be exported to begin with, but I don't have a strong opinion either way
fyi Vincent's opinion on this @jrieke @sfc-gh-jcarroll |
|
My understanding is we only saw this a couple times in the wild and it's easy to fix in a backwards compatible way inside of a given component or app. Assuming that's true, I'm also fine to NOT revive this behavior and not merge this (and I think I have a slight preference to same). We should just make sure we fix the instances like these in the docs that refer to this now-not-functioning behavior. Unless someone else sees a good reason to keep it. cc @sfc-gh-dmatthews thoughts? |
|
This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. |
Describe your changes
Closes #8644
From this comment:
The usage of
components.htmlviast.components.v1.htmlwas possible prior to1.34due to this import:from streamlit.components.v1.components import ComponentRegistry, which made it transiently available. This means that the usage viast.components.v1.htmlis only possible whenserver.pyis imported at one point, which is not the case for example when running Streamlit as a bare-script.We have discussed this and thought about bringing the behavior back, but this time explicitly and with tests 🙂
Final decision is still pending.
Testing Plan
Contribution License Agreement
By submitting this pull request you agree that all contributions to this project are made under the Apache 2.0 license.