Search regression data for "End of data reached." and improve buffer length detection
drewnoakes opened this issue · comments
Our reader classes validate range arguments. They throw IOException for attempts to read beyond the allotted range. See here for an example.
Exceptions can tunnel a long way before being caught. By replacing these exceptions with graceful error handling, we might proceed to extract more metadata successfully.
We should search the regression suite's output files for occurrences of the error message:
End of data reached.
I believe several, if not most, of these could be identified with a quick range calculation ahead of read operations.
Also messages for out-of-bounds errors.