You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
perf: avoid copying large bytearrays when binary output is summarized #602
For an already-large bytearray, bytes(value) allocates a second full buffer before the code discovers that the value is too big to inline. The configured limit defaults to only 256 bytes, so this copy is unnecessary on the summarized branch.
This is a source observation, not a live CUBRID memory benchmark.
Bounded optimization
Check len(value)before converting/copying. For above-limit binary, return the same <binary N bytes> summary without a bytes() conversion. Preserve base64 output for small binary and the max_inline_binary_bytes=0 case.
Add offline tests for bytes and bytearray at lengths 0, the configured boundary, and one larger; confirm correct output and no eager copy on large bytearray via a suitable spy/instrumented test object.
Activity