[Bugfix] Updates "un-committed" when rebasing #17430
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Adds a failing test case where an update that was committed is later skipped over during a rebase. This should never happen.
|
This pull request is automatically built and testable in CodeSandbox. To see build info of the built libraries, click here or the icon next to each commit SHA. Latest deployment of this branch, based on commit 97ec757:
|
Details of bundled changes.Comparing: 54f6673...97ec757 react-reconciler
react-dom
react-art
react-test-renderer
react-native-renderer
ReactDOM: size: 🔺+0.2%, gzip: 🔺+0.3% Size changes (stable) |
Details of bundled changes.Comparing: 54f6673...97ec757 react-dom
react-native-renderer
react-test-renderer
react-art
react-reconciler
Size changes (experimental) |
acdlite
force-pushed
the
never-uncommit
branch
from
4a1e6eb to
f8adc01
Compare
acdlite
changed the title
To enforce that updates that are committed can never be un-committed, even during a rebase, we need to track which updates are part of the rebase. The current implementation doesn't do this properly. It has a hidden assumption that, when rebasing, the next `renderExpirationTime` will represent a later expiration time that the original one. That's usually true, but it's not *always* true: there could be another update at even higher priority. This requires two extra fields on the update queue. I really don't like that the update queue logic has gotten even more complicated. It's tempting to say that we should remove rebasing entirely, and that update queues must always be updated at the same priority. But I'm hesitant to jump to that conclusion — rebasing can be confusing in the edge cases, but it's also convenient. Enforcing single-priority queues would really hurt the ergonomics of useReducer, for example, where multiple values are centralized in a single queue. It especially hurts the ergonomics of classes, where there's only a single queue per class. So it's something to think about, but I don't think "no more rebasing" is an acceptable answer on its own.
acdlite
force-pushed
the
never-uncommit
branch
from
b4f4779 to
ac040cc
Compare
acdlite
commented
Nov 26, 2019
| newBaseUpdate = prevUpdate; | ||
| newBaseState = newState; | ||
| newRebaseTime = renderExpirationTime; |
There was a problem hiding this comment.
Ooh I think this should be the lower of renderExpirationTime and the old rebaseTime. In case there are two rebases in a row. I'll add a new test.
There was a problem hiding this comment.
It would also have to be the further of the two RebaseEnd then, no?
There was a problem hiding this comment.
Actually I don't think this will work. There can be arbitrary numbers of rebases, each with their own ranges and expiration times.
Will need to reevaluate.
sebmarkbage
approved these changes
Nov 26, 2019
|
The approach here won't work because of #17430 (comment) Closing in favor of #17483 stack |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
To enforce that updates that are committed can never be un-committed, even during a rebase, we need to track which updates are part of the rebase. The current implementation doesn't do this properly. It has a hidden assumption that, when rebasing, the next
renderExpirationTimewill represent a later expiration time that the original one. That's usually true, but it's not always true: there could be another update at even higher priority.This requires two extra fields on the update queue. I really don't like that the update queue logic has gotten even more complicated. It's tempting to say that we should remove rebasing entirely, and that update queues must always be updated at the same priority. But I'm hesitant to jump to that conclusion — rebasing can be confusing in the edge cases, but it's also convenient. Enforcing single-priority queues would really hurt the ergonomics of useReducer, for example, where multiple values are centralized in a single queue. It especially hurts the ergonomics of classes, where there's only a single queue per class.
So it's something to think about, but I don't think "no more rebasing" is an acceptable answer on its own.