model-writtenrun 8537 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Zheng Hacker reported that loread() lengths near INT_MAX overflow the allocation-size calculation and produce a confusing error, suggesting either clamping the length or explicitly rejecting it in #1.
- Daniel Gustafsson chose explicit rejection consistent with lo_get() and posted a patch checking the allocation limit before palloc() in #2.
- Tom Lane questioned whether improving this harmless error justified translation work in #3; Daniel explained that the patch reused an existing large-object error message and agreed that the existing behavior posed no apparent risk in #4.
- Tom accepted the patch provided it remained master-only in #5, and Daniel agreed that backpatching was unwarranted in the thread's final message, #6.
- Daniel subsequently committed the patch from #2 without code changes as 2c077c7d60, crediting Tom's review and documenting that the safe existing behavior did not warrant backpatching.