AynSiddiqui commented on issue #77: URL: https://github.com/apache/arrow-go/issues/77#issuecomment-6768060924
I tried the stream corpus test with a checked allocator on `66722e30`: 35 of 84 inputs leave allocations after reader release. Moving the slice evaluation before `Retain` in `memory.SliceBuffer` reduces that to 3 failing inputs. Those expose additional cleanup gaps in `loadList`, `loadStruct` and the pre-1.0 union-validity check. I also verified that PyArrow 26.0.1 rejects both inputs Go accepts: `clusterfuzz-testcase-minimized-arrow-ipc-stream-fuzz-4757582821064704` (`Invalid flatbuffers message`) and `crash-5e88bae6ac5250714e8c8bc73b9d67b949fadbb4` (decimal precision 0). Would a small `SliceBuffer` fix with a unit test be a useful first PR, followed by IPC cleanup together with the corpus test? Should I track the two validation differences separately? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
