msglist test: Fix a state leak from mutating eg.selfUser #1713
Conversation
This file contains hidden or 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
Fixes zulip#1712. Our CI has been failing in this test file since yesterday; this change fixes that. This test had been having the store handle a RealmUserUpdateEvent affecting `eg.selfUser`. That means the store mutates the actual User object it has... which, in this test context, means the value that `eg.selfUser` is bound to as a final variable. As a result, when the test gets cleaned up with testBinding.reset, the store gets discarded as usual so that the next test will get a fresh store... but the User object at `eg.selfUser` is still the one that's been mutated by this test. That's buggy in principle. Concretely, here, it causes the self-user to have a non-null avatar. When a later test sends a message and causes an outbox-message to appear in the tree, that results in a NetworkImage. And that throws an error, because no HttpClient provider has been set, because that latter test wasn't expecting to create any `NetworkImage`s. That's the failure we've been seeing in CI. It's still a bit mysterious why this had previously been working (or anyway these tests hadn't been failing). It started failing with an upstream change merged yesterday, which makes changes to NetworkImage that look innocuous: its `==` and `hashCode` become finer-grained, and its `toString` more detailed. In any case, the bug is ours to fix. It'd also be good to follow up here by systematically preventing this sort of state leak between tests. But that comes after getting our CI passing again.
gnprice
added a commit
to gnprice/zulip-flutter
that referenced
this pull request
Jul 18, 2025
This is the same sort of state leak that caused zulip#1712; this instance of it just hasn't happened to break any tests for us yet. We'll soon arrange things so that this sort of state-leaking mutation causes an immediate error. This is one of the three total places where it turns out we had such mutations, including the one we just fixed in a04b44e (zulip#1713). The second of these tests ("no error if recipient was deactivated …") wasn't actually mutating the shared example user `eg.otherUser`, because secretly `setupToMessageActionSheet` makes a new User object with the same user ID and puts that in the store. Still, it *looked* like it was; best to do something that clearly looks correct instead. The first of these tests was indeed mutating `eg.selfUser`, just as it looks like it's doing.
chrisbobbe
pushed a commit
that referenced
this pull request
Jul 22, 2025
This is the same sort of state leak that caused #1712; this instance of it just hasn't happened to break any tests for us yet. We'll soon arrange things so that this sort of state-leaking mutation causes an immediate error. This is one of the three total places where it turns out we had such mutations, including the one we just fixed in a04b44e (#1713). The second of these tests ("no error if recipient was deactivated …") wasn't actually mutating the shared example user `eg.otherUser`, because secretly `setupToMessageActionSheet` makes a new User object with the same user ID and puts that in the store. Still, it *looked* like it was; best to do something that clearly looks correct instead. The first of these tests was indeed mutating `eg.selfUser`, just as it looks like it's doing.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Fixes #1712.
Our CI has been failing in this test file since yesterday; this change fixes that.
This test had been having the store handle a RealmUserUpdateEvent affecting
eg.selfUser. That means the store mutates the actual User object it has... which, in this test context, means the value thateg.selfUseris bound to as a final variable.As a result, when the test gets cleaned up with testBinding.reset, the store gets discarded as usual so that the next test will get a fresh store... but the User object at
eg.selfUseris still the one that's been mutated by this test.That's buggy in principle. Concretely, here, it causes the self-user to have a non-null avatar. When a later test sends a message and causes an outbox-message to appear in the tree, that results in a NetworkImage. And that throws an error, because no HttpClient provider has been set, because that latter test wasn't expecting to create any
NetworkImages. That's the failure we've been seeing in CI.It's still a bit mysterious why this had previously been working (or anyway these tests hadn't been failing). It started failing with an upstream change merged yesterday, which makes changes to NetworkImage that look innocuous: its
==andhashCodebecomefiner-grained, and its
toStringmore detailed. In any case,the bug is ours to fix.
It'd also be good to follow up here by systematically preventing this sort of state leak between tests. But that comes after getting our CI passing again.