-
-
Notifications
You must be signed in to change notification settings - Fork 32.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[styles] Improve ref forwarding #13676
[styles] Improve ref forwarding #13676
Conversation
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@oliviertassinari Currently things like |
@eps1lon How do you want to move forward? |
|
Sounds good, it's large enterprise, thank you for leading it! Do we still need |
I think we can keep it if users do something like |
Alright the issues described by @oliviertassinari are now solved. Only caveat is a deprecation warning in our docs now for |
- It forwards *non React static* properties so this HOC is more "transparent". | ||
- It forwards refs to the inner component. | ||
- The `innerRef` prop is deprecated. Use `ref` instead. | ||
- It does **not** copy over statics. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Forgotten in #13698
const ThemedDiv = withTheme()('div'); | ||
|
||
mount( | ||
<> |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Fragment?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was an issue with refs on root components when mounting in enzyme. I think I added a comment somewhere but this might've been on a local branch.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@oliviertassinari I can't find the original issue unfortunately. However it does indeed break if I remove the fragment. ref.current
will be undefined
.
@material-ui/styles: parsed: -3.06% 😍, gzip: -2.18% 😍 Details of bundle changes.Comparing: 8d46415...ff974b3
|
3ea3061
to
4998d69
Compare
@@ -1,4 +1,3 @@ | |||
/* istanbul ignore file */ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm removing all the istanbul ignore, @eps1lon is right, let's leave the 100% utopia.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
To be fair that broken window analogy you linked made total sense to me. I just think that these comments won't help in that regard. At least this way we have the actual number to look at. Previously it was 100% (kind of; who knows how good it is actually). Now we know how good/bad we are and have tools helping us analyze it. Codesearch is not really that helpful IMO. It's like prefering // TODO
over creating issues.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I agree.
'Material-UI: The `innerRef` prop is deprecated and will be removed in v5. ' + | ||
'Refs are now automatically forwarded to the inner component.', | ||
); | ||
// return new Error( |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's raising a warning as the ButtonBase is still using this API.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Right. I think I annotated it with @deprecated
somewhere to indicate a "soft deprecation". In a perfect world the eslint rule would detect these. I don't like runtime deprecation warnings that have the same level as any other warning (hello ReactDOM.findDOMNode
). A lint rule makes more sense to me.
@@ -51,9 +51,9 @@ describe('withStyles', () => { | |||
|
|||
const ref = React.createRef(); | |||
mount( | |||
<> | |||
<React.Fragment> |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What should we do with the Fragment. Should we embrace <>
everywhere, should only use React.Fragment or it doesn't matter 🤔. I'm reducing entropy until we decide.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think React.Fragment
is required if fragment refs land. So maybe enforce that?
I want to merge this pull request so I can finally complete #14560! |
Breaking change
withStyles
andwithTheme
now forward their refs to the decorated component:ref
on the wrapped component to get a DOM node viaref
useRootRef
instead if you have no control over the componentor
React.forwardRef
if you do have control:innerRef
is being deprecated. It is however prioritized overref
:@material-ui/styles
is still in alpha so the breaking change should not require a major version bump. It will be however breaking since we will use@material-ui/styles
in v4 over@material-ui/core/styles
.