Fix Image.network not using cache when headers are specified #176831
Conversation
There was a problem hiding this comment.
Code Review
This pull request corrects the hashCode implementation for NetworkImage in both its IO and web versions. The change addresses a bug where the hash code for the headers map was computed based on object identity, leading to cache misses for Image.network when new but identical header maps were used. By using MapEquality from the collection package, the hash code is now correctly based on the map's contents, aligning it with the operator== implementation. A new test is also added to verify that NetworkImage with identical arguments correctly utilizes the image cache, preventing future regressions.
cb3ec5b to
dd58403
Compare
There was a problem hiding this comment.
Thanks @rajveermalviya for the diagnosis and the fix! These changes look good; just nits below on the test.
I also tweaked the description (which becomes the merged commit message) to point to the previous PR as well as the commit, for ease of cross-reference.
| // NetworkImage should not make another request. | ||
| mockHttpClient.thrownError = Error(); |
There was a problem hiding this comment.
This line doesn't totally reassure me that the test will notice if the NetworkImage does make another request. With the current implementation of _FakeHttpClient, that holds because thrownError takes precedence over request; but there's no reason it has to be that way, and someone could well refactor that in the future without noticing that it causes this test to be ineffective.
Let's instead put this in the same terms as the .request.response setup above, so that it clearly overrides it:
| // NetworkImage should not make another request. | |
| mockHttpClient.thrownError = Error(); | |
| // NetworkImage should not make another request. | |
| mockHttpClient.request.response.statusCode = HttpStatus.badRequest; |
Alternatively, could set debugNetworkImageHttpClientProvider to a fresh _FakeHttpClient instance.
| class _FakeHttpClientResponse extends Fake implements HttpClientResponse { | ||
| bool drained = false; | ||
|
|
||
| @override | ||
| final HttpHeaders headers = _FakeHttpHeaders(); |
There was a problem hiding this comment.
This change isn't needed, is it? The new test adds headers only in the request, not the response.
Fix a bug introduced by 97aae2a, where `hashCode` is giving different results for values which `==` says are equal. Which broke caching for uses of `Image.network` widget when `headers` map was passed.
dd58403 to
6efc307
Compare
|
Thanks for the review @gnprice! Addressed those comments. |
There was a problem hiding this comment.
Thanks! These changes LGTM.
BTW for future PRs in this repo, the most convenient thing is to push revisions as added commits on top of the existing ones. (The exception is when there's a need to rebase atop a newer main.)
That's helpful for later reviewers or other readers to be able to follow the thread in the GitHub web interface, because it effectively keeps the full history of your changes visible here. On merge, the changes will get squashed into one commit anyway, so there's no harm to the clarity of the final history in main.
Fix a bug introduced by 97aae2a (#171916), where
hashCodeis giving different results for values which==says are equal. Which broke caching for uses ofImage.networkwidget whenheadersmap was passed.Pre-launch Checklist
///).If you need help, consider asking for advice on the #hackers-new channel on Discord.
Note: The Flutter team is currently trialing the use of Gemini Code Assist for GitHub. Comments from the
gemini-code-assistbot should not be taken as authoritative feedback from the Flutter team. If you find its comments useful you can update your code accordingly, but if you are unsure or disagree with the feedback, please feel free to wait for a Flutter team member's review for guidance on which automated comments should be addressed.