Move connection hook to improve performance #143
Conversation
This might not be the best way to handle this, in particular the local PV is now limited but it does work for the majority of cases I am interested in testing at the moment
This reverts commit e25a569.
|
This PR currently prevents us from calculating rules at runtime, but we can't do this effectively anyway for many props (such as position, font etc). I think this should be looked at as a separate concern. |
Required mocking out multiple modules which can be reasonably painful. Some improvements could perhaps be made to the implementation
| containerProps: any & { id: string }; | ||
| widgetProps: any; |
There was a problem hiding this comment.
I would like to stop these allowing any through in the future
|
All readbacks looking at the same PV. Clear speedup from hooked and wrapped components changes. Performance comparable to dbwr for this test. |
|
Interesting that this branch improves performance, as it seems a fairly minor change. As the performance is comparable to what you called 'hooked-components' I think this one is more useful. |
The connection hook is required at runtime, but many of the other hooks (useMacros etc) are not. Therefore, this PR moves the connection hook into a separate widget which is rendered after the load-time hooks in order to separate concerns. This appears to have provided an approximately 36% increase in speed over what is currently on the master branch when measuring both with production build and the profile flag (
npm run build -- --profile)It also does not disrupt our current widget layout but does give us a method for concerning load-time and runtime concerns.