Fix fetch response stream cancellation

[simolus3]

The fetch client turns a ReadableStreamDefaultReader into a Dart Stream by repeatedly invoking reader.read() in an async* function.

This generally works well, but a problem with async* functions is that they can only respect cancellations when yielding. This results in cancellations taking longer than they need to in the following case:

1. A chunk of data is reported from the reader.
2. This is emitted by the stream, and we go into the next iteration of the _readBody loop.
3. Asynchronously, the listener calls .cancel() on its subscription.
4. This now has to wait for the next chunk of the source stream instead of aborting the request.

This fixes the issue by rewriting the logic to use a StreamController directly.

Closes #1820.

cc @brianquinlan

---

• I’ve reviewed the contributor guide and applied the relevant portions to this PR.

Contribution guidelines:

• See our contributor guide for general expectations for PRs.
• Larger or significant changes should be discussed in an issue before creating a PR.
• Contributions to our repos should follow the Dart style guide and use dart format.
• Most changes should add an entry to the changelog and may need to rev the pubspec package version.
• Changes to packages require corresponding tests.

Note that many Dart repos have a weekly cadence for reviewing PRs - please allow for some latency before initial review feedback.

[github-actions]

Package publishing

Package	Version	Status	Publish tag (post-merge)
package:cronet_http	1.5.0	already published at pub.dev
package:cupertino_http	2.4.0-wip	WIP (no publish necessary)
package:http	1.6.0-wip	WIP (no publish necessary)
package:http2	3.0.0	ready to publish	http2-v3.0.0
package:http_multi_server	3.2.2	already published at pub.dev
package:http_parser	4.1.2	already published at pub.dev
package:http_profile	0.1.1-wip	WIP (no publish necessary)
package:ok_http	0.1.1-wip	WIP (no publish necessary)
package:web_socket	1.0.1	already published at pub.dev
package:web_socket_channel	3.0.3	already published at pub.dev

Documentation at https://github.com/dart-lang/ecosystem/wiki/Publishing-automation.

[github-actions]

PR Health

License Headers ✔️

// Copyright (c) 2025, the Dart project authors. Please see the AUTHORS file
// for details. All rights reserved. Use of this source code is governed by a
// BSD-style license that can be found in the LICENSE file.

Files
no missing headers

All source files should start with a license header.

Unrelated files missing license headers

Files
pkgs/http/example/main.dart

This check can be disabled by tagging the PR with skip-license-check.

API leaks ✔️

The following packages contain symbols visible in the public API, but not exported by the library. Export these symbols or remove them from your publicly visible API.

Package	Leaked API symbol	Leaking sources

This check can be disabled by tagging the PR with skip-leaking-check.

Breaking changes ✔️

Package	Change	Current Version	New Version	Needed Version	Looking good?
http	None	1.5.0	1.6.0-wip	1.5.0	✔️

This check can be disabled by tagging the PR with skip-breaking-check.

Coverage ⚠️

File	Coverage
pkgs/http/lib/src/browser_client.dart	💔 Not covered

This check for test coverage is informational (issues shown here will not fail the PR).

This check can be disabled by tagging the PR with skip-coverage-check.

Changelog Entry ✔️

Package	Changed Files

Changes to files need to be accounted for in their respective changelogs.

This check can be disabled by tagging the PR with skip-changelog-check.

[brianquinlan]

Very high quality PR! Especially the added conformance test (a few of which seem to be failing).

[simolus3]

It looks like the remaining failures are all infra or appear to be unrelated to my changes. Let me know if there's anything I should look into.

[brianquinlan]

It looks like the remaining failures are all infra or appear to be unrelated to my changes. Let me know if there's anything I should look into.

Yep, these have nothing to do with your change.

[brianquinlan]

Hey @simolus3 ,

I just wanted to let you know what I'm still looking at your change. I've included a few nit picky comments but overall its excellent. What I don't like is the increased complexity of this client. But I can't think of a better way of doing this so probably it is fine the way it is. But it might take me a few days to convince myself of that ;-)

[lrhn]

I think it's pretty good. I can simplify it a little, but the underlying idea is the same.

[simolus3]

Alright, I have adopted most of the suggestions shared by @lrhn. I think they make the control flow a lot more obvious 👍

I think the correct behavior (which we also had with the old async* implementation) is to only call getReader() when the stream is actually listened to, so I've moved most of the setup into the onListen callback. That also avoids the use of finalizers here.

Despite being a single-use stream, I think using Stream.multi may be helpful here to be explicit about adding events synchronously? I've also added a comment on each addSync / addErrorSync / closeSync invocation to explain why it's safe.

[brianquinlan]

It looks like one test is failing on WASM.

[simolus3]

Thanks, I've fixed that test.

[brianquinlan]

It looks like packages cupertino_http and cronet_http both fail on "cancel paused stream via abortable request". Let me see if I can figure out why...

[brianquinlan]

@simolus3 Thanks for sticking with this and for the great work!

[brianquinlan]

The remaining CI failures are unrelated.

[simolus3]

Thanks for fixing the test and reviewing this!