Skip to content
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

[core] add AsyncControllableTextArea based on AsyncControllableInput & add async control option for TextArea component #6312

Merged
merged 13 commits into from
Aug 23, 2023

Conversation

nurseiit
Copy link
Member

@nurseiit nurseiit commented Aug 1, 2023

Related to #4678.

Checklist

  • Includes tests
  • Update documentation

Changes proposed in this pull request:

a. (Extends the existing AsyncControllableInput with an option to render a textarea element. (the type of Props is quite complicated now, and I am open to suggestions to further simplify it)) -- initial approach. Ended up with a useAsyncControllableValue hook to use in AsyncControllableTextArea component.

b. Adds asyncControl option to TextArea component that reuses AsyncControllableInput uses AsyncControllableTextArea when set to true.

Reviewers should focus on:

Basically the only thing I was afraid of and tried to avoid is any external incompatibilities with the API and component props types. But since AsyncControllableInput is not exported externally and only used as a part of other components I think we can get away with the introduced complex types.

Also I generalised the tests for AsyncControllableInput instead of copying them, one can argue this was unnecessary, but I am also open for any suggestions on this.

Thanks!

@adidahiya
Copy link
Contributor

add correct types & copy tests from default input

Build artifact links for this commit: documentation | landing | table | demo

This is an automated comment from the deploy-preview CircleCI job.

@adidahiya
Copy link
Contributor

fix lint, object order

Build artifact links for this commit: documentation | landing | table | demo

This is an automated comment from the deploy-preview CircleCI job.

Comment on lines 34 to 37
// NOTE: these are copied from the React.HTMLAttributes interface definition.
onChange?: React.ChangeEventHandler<AsyncControllableElement<T>> | undefined;
onCompositionStart?: React.CompositionEventHandler<AsyncControllableElement<T>> | undefined;
onCompositionEnd?: React.CompositionEventHandler<AsyncControllableElement<T>> | undefined;
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

not sure if this is the best way to do it, but I would argue this is fine since this component is not exported directly.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I did not want to further complicate the types by excluding and extending the native interfaces above.

@nurseiit nurseiit changed the title WIP: async control option for TextArea AsyncControllableInput to render textarea element & async control option for TextArea component Aug 2, 2023
@nurseiit nurseiit marked this pull request as ready for review August 2, 2023 13:09
@adidahiya
Copy link
Contributor

refactor tests

Build artifact links for this commit: documentation | landing | table | demo

This is an automated comment from the deploy-preview CircleCI job.

@nurseiit nurseiit requested a review from adidahiya August 2, 2023 13:11
@nurseiit nurseiit changed the title AsyncControllableInput to render textarea element & async control option for TextArea component [core] AsyncControllableInput to render textarea element & async control option for TextArea component Aug 2, 2023
@gluxon gluxon self-assigned this Aug 8, 2023
@nurseiit
Copy link
Member Author

Hi @gluxon, any chance you could take a look a this? Thanks!

Copy link
Contributor

@gluxon gluxon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Apologies for the late response — It took me a bit of time to come around to it, but I like how this takes advantage of generic types to ensure type safety of the input props.


A discussion of a long-term approach may make sense. Zooming out a bit, more recent React code tends to lean on hooks for component logic sharing. I think the need to add additional functionality to this existing AsyncControllableInput class component comes from the fact that it was written when component composition was a more common logic reuse pattern. In that prior world, this PR makes sense.

If we were to imagine an alternative hooks-based approach, I think adding support for new HTML textual input elements becomes significantly easier. This example isn't complete and likely fails to compile, but here's the rough direction I was thinking:

function AsyncControllableInput(props: P) {
    const { value, onChange, ...rest } = props;
    const { textInputProps } = useAsyncControlledValue({ value, onChange });

    return <input {...rest} {...textInputProps} />
}

function AsyncControllableTextarea(props: P) {
    const { value, onChange, ...rest } = props;
    const { textInputProps } = useAsyncControlledValue({ value, onChange });

    return <textarea {...rest} {...textInputProps} />
}

function useAsyncControllableValue(opts) {
    const [isComposing, setIsComposing] = React.useState(false);
    const [nextValue, setNextValue ] = React.useState(/* ... */);
    const [hasPendingUpdate, setHasPendingUpdate ] = React.useState(/* ... */);

    function onCompositionStart() {
        setIsComposing(true)
    }

    function onCompositionEnd() {
        setIsComposing(false)
    }

    const value = isComposing || hasPendingUpdate ? nextValue : value

    /* ... */

    return {
        value,
        onChange,
        onCompositionEnd,
        onCompositionStart
    };
}

Since this PR adds a new functionality, I do wonder if it makes sense to create a new AsyncControllableTextarea that's based upon the hook approach. We would then battle-test that purely new functionality before switching AsyncControllableInput to be implemented from the same hook in a future Blueprint release.

I'm also okay extending the existing AsyncControllableInput to be generic since it's private anyway. I do think the hook approach is the preferable long-term approach though. What do you think?

Basically the only thing I was afraid of and tried to avoid is any external incompatibilities with the API and component props types. But since AsyncControllableInput is not exported externally and only used as a part of other components I think we can get away with the introduced complex types.

I appreciate that! I scanned through the PR looking for changes to external APIs and don't see any other than the new asyncControl prop that was intended.

Comment on lines 213 to 215
describe("<AsyncControllableInput>", () => {
runAsyncControllableInputTests({ tagName: "input", inputType: "text" });
runAsyncControllableInputTests({ tagName: "textarea" });
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I like the test deduplication here. Thanks for doing that.

What are your thoughts on describe.each? That would be another way to write this in more idiomatic Jest. https://jestjs.io/docs/api#describeeachtablename-fn-timeout

@@ -170,13 +190,22 @@ export class TextArea extends AbstractPureComponent<TextAreaProps, TextAreaState
};
}

return (
return asyncControl ? (
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this seem to work when testing live? The code changes look correct. Just wondering if <TextArea asyncControl /> works as expected.

@nurseiit
Copy link
Member Author

Apologies for the late response — It took me a bit of time to come around to it, but I like how this takes advantage of generic types to ensure type safety of the input props.

A discussion of a long-term approach may make sense. Zooming out a bit, more recent React code tends to lean on hooks for component logic sharing. I think the need to add additional functionality to this existing AsyncControllableInput class component comes from the fact that it was written when component composition was a more common logic reuse pattern. In that prior world, this PR makes sense.

If we were to imagine an alternative hooks-based approach, I think adding support for new HTML textual input elements becomes significantly easier. This example isn't complete and likely fails to compile, but here's the rough direction I was thinking:

function AsyncControllableInput(props: P) {
    const { value, onChange, ...rest } = props;
    const { textInputProps } = useAsyncControlledValue({ value, onChange });

    return <input {...rest} {...textInputProps} />
}

function AsyncControllableTextarea(props: P) {
    const { value, onChange, ...rest } = props;
    const { textInputProps } = useAsyncControlledValue({ value, onChange });

    return <textarea {...rest} {...textInputProps} />
}

function useAsyncControllableValue(opts) {
    const [isComposing, setIsComposing] = React.useState(false);
    const [nextValue, setNextValue ] = React.useState(/* ... */);
    const [hasPendingUpdate, setHasPendingUpdate ] = React.useState(/* ... */);

    function onCompositionStart() {
        setIsComposing(true)
    }

    function onCompositionEnd() {
        setIsComposing(false)
    }

    const value = isComposing || hasPendingUpdate ? nextValue : value

    /* ... */

    return {
        value,
        onChange,
        onCompositionEnd,
        onCompositionStart
    };
}

Since this PR adds a new functionality, I do wonder if it makes sense to create a new AsyncControllableTextarea that's based upon the hook approach. We would then battle-test that purely new functionality before switching AsyncControllableInput to be implemented from the same hook in a future Blueprint release.

I'm also okay extending the existing AsyncControllableInput to be generic since it's private anyway. I do think the hook approach is the preferable long-term approach though. What do you think?

Basically the only thing I was afraid of and tried to avoid is any external incompatibilities with the API and component props types. But since AsyncControllableInput is not exported externally and only used as a part of other components I think we can get away with the introduced complex types.

I appreciate that! I scanned through the PR looking for changes to external APIs and don't see any other than the new asyncControl prop that was intended.

I really do like the hooks approach! That's what I was aiming for at first, but then gave up unfortunately. In any case, I will revisit this again with hooks. Thanks for reviewing!

@gluxon
Copy link
Contributor

gluxon commented Aug 17, 2023

I really do like the hooks approach! That's what I was aiming for at first, but then gave up unfortunately. In any case, I will revisit this again with hooks. Thanks for reviewing!

Please reach out if you want a second pair of eyes while looking at this! 🙂

@nurseiit
Copy link
Member Author

I really do like the hooks approach! That's what I was aiming for at first, but then gave up unfortunately. In any case, I will revisit this again with hooks. Thanks for reviewing!

Please reach out if you want a second pair of eyes while looking at this! 🙂

Thanks! I updated the PR with hooks alternative based on the AsyncControllableInput component and your comment above. Let me know what do you think 😌

@adidahiya
Copy link
Contributor

add more comments

Build artifact links for this commit: documentation | landing | table | demo

This is an automated comment from the deploy-preview CircleCI job.

Copy link
Contributor

@gluxon gluxon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for working on this! I personally like how this came out. Just have a few more questions and I think we'll be ready to merge soon.

Played around with the latest hook-based version live. Seeing all the expected expected state updates.

Screenshot 2023-08-19 at 1 22 50 AM

packages/core/src/components/forms/textArea.tsx Outdated Show resolved Hide resolved
Comment on lines 96 to 103
if (propValue === nextValue) {
// parent has processed and accepted our update
if (hasPendingUpdate) {
setValue(propValue);
setHasPendingUpdate(false);
} else if (value !== nextValue) {
setValue(nextValue);
}
Copy link
Contributor

@gluxon gluxon Aug 19, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There might be opportunity to simplify this block. I know it was copied from the original version, but since we're rewriting it anyway, it might be cool to revisit.

Staring at this part, I think I can convince myself that these blocks are equivalent:

if (propValue === nextValue) {
    // parent has processed and accepted our update
    if (hasPendingUpdate) {
        setValue(propValue);
        setHasPendingUpdate(false);
    } else if (value !== nextValue) {
        setValue(nextValue);
    }
} else {
if (propValue === nextValue) {
    // parent has processed and accepted our update
    setValue(propValue);
    setHasPendingUpdate(false);
} else {

The theory being that

  1. propValue and nextValue are the same, so either can be used to setValue.
  2. React should no-op the setHasPendingUpdate(false) call if it's already false.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Taking this a bit further, I think the nested logic can be much flatter.

if (!isStateDerivationPaused) {
    const userTriggeredUpdate = nextValue !== value;

    if (userTriggeredUpdate && propValue === nextValue) {
        // parent has processed and accepted our update
        setValue(propValue);
        setHasPendingUpdate(false);
    } else if (userTriggeredUpdate && propValue === value) {
        // we have sent the update to our parent, but it has not been processed yet. just wait.
        // DO NOT set nextValue here, since that will temporarily render a potentially stale controlled value,
        // causing the cursor to jump once the new value is accepted
        if (!hasPendingUpdate) {
            setHasPendingUpdate(true);
        }
    } else if (userTriggeredUpdate || value !== propValue || hasPendingUpdate) {
        // accept controlled update, could be confirming or denying user action
        setValue(propValue);
        setNextValue(propValue);
        setHasPendingUpdate(false);
    }
}

But I could definitely use help making sure I refactored that correctly. The general strategy was to propagate the userTriggeredUpdate condition down and then remove layers of nesting.

If you would rather not think about further refactors and prefer to keep the existing structure as-is to match the current getDerivedStateFromProps approach, I think that's okay too!

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I refactored this bit to be flatter and (arguably) more readable than the initial getDerivedStateFromProps version in the latest commit. Thanks for the ideas above!

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good!

@nurseiit
Copy link
Member Author

Thanks for the review @gluxon! Updated based on the latest comments / suggestions 👍

@adidahiya
Copy link
Contributor

flatten & simplify state derival + comments

Build artifact links for this commit: documentation | landing | table | demo

This is an automated comment from the deploy-preview CircleCI job.

Copy link
Contributor

@gluxon gluxon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the updates! I think everything looks good. Expecting to approve and merge once the discussion about tests https://github.com/palantir/blueprint/pull/6312/files/bdc07b6e4a19ef154d84cbc233943985c6f91cbd#r1300068035 wraps up.

@nurseiit
Copy link
Member Author

Updated the tests & added a clarifying note at the top of the test file ☺️ Thanks for reviewing this @gluxon!

@nurseiit nurseiit changed the title [core] AsyncControllableInput to render textarea element & async control option for TextArea component [core] add AsyncControllableTextarea based on AsyncControllableInput & add async control option for TextArea component Aug 22, 2023
@nurseiit nurseiit changed the title [core] add AsyncControllableTextarea based on AsyncControllableInput & add async control option for TextArea component [core] add AsyncControllableTextArea based on AsyncControllableInput & add async control option for TextArea component Aug 22, 2023
@adidahiya
Copy link
Contributor

dedup tests & add note

Build artifact links for this commit: documentation | landing | table | demo

This is an automated comment from the deploy-preview CircleCI job.

Copy link
Contributor

@gluxon gluxon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you for the improvements and new feature @nurseiit! 🎉

@gluxon gluxon merged commit 55f525d into develop Aug 23, 2023
@gluxon gluxon deleted the na/async-control-textarea branch August 23, 2023 16:41
Copy link
Contributor

@adidahiya adidahiya left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks @nurseiit & @gluxon! I like the new approach of using the hook to implement this behavior. We should be able to migrate AsyncControllableInput to use it as well soon.

I can fix my minor review comments in a follow-up PR.


import * as React from "react";

interface IUseAsyncControllableValueProps<E extends HTMLInputElement | HTMLTextAreaElement> {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: same comment about avoiding the I prefix

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants