| commit | d210241eb8433bf569ac853428e7390f6aeab8be | |
| author | Andrew Hilger <[email protected]> | |
| Fri, 31 Oct 2025 00:18:41 +0000 (30 17:18 -0700) | ||
| committer | meta-codesync[bot] <215208954+meta-codesync[bot]@users.noreply.github.com> | |
| Fri, 31 Oct 2025 00:18:41 +0000 (30 17:18 -0700) | ||
| tree | 88f0223dcd89eb83c03567d1df6ab795b230208b | treesnapshot (tar.gz zip) |
| parent | 995dad4c43b3c67146be883079f615a4dfe31921 | commitdiff |
stress tests for incomplete stream consume
Summary:
Can repro crash if the stream is not fully consumed. A few ways to trigger:
1. don't consume anything from stream after starting `sink`.
2. only partially consume the stream
The following crash is unfixed:
- don't even call the `ClientSink.sink` method (see next diff)
Another way to trigger crash appears to be double calling the stream / sink. Want to isolate that a bit more.
Reviewed By: prakashgayasen
Differential Revision: D85624468
fbshipit-source-id: aba0b2335528bebcbf2692741d45db5f80204add
Summary:
Can repro crash if the stream is not fully consumed. A few ways to trigger:
1. don't consume anything from stream after starting `sink`.
2. only partially consume the stream
The following crash is unfixed:
- don't even call the `ClientSink.sink` method (see next diff)
Another way to trigger crash appears to be double calling the stream / sink. Want to isolate that a bit more.
Reviewed By: prakashgayasen
Differential Revision: D85624468
fbshipit-source-id: aba0b2335528bebcbf2692741d45db5f80204add
| third-party/thrift/src/thrift/lib/python/test/bidi.py | diffblobblamehistory |