model-writtenrun 3624 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Peter Eisentraut proposed clearer command assembly using extensible buffers in initdb, pg_dumpall, and pg_regress, leaving pg_ctl out because its implementation differed substantially #1.
- Heikki Linnakangas endorsed the cleanup but preferred StringInfo's exit-on-allocation-failure behavior over PQExpBuffer's empty-buffer behavior, without requiring that broader issue to be fixed immediately #2.
- Peter Eisentraut committed the original patch unchanged as cccdbc5d95 and agreed to address allocation-failure behavior separately #3.
- Seven months later, Peter Eisentraut submitted a new series converting the regression and isolation test drivers to StringInfo and proposing a separate pg_upgrade conversion to PQExpBuffer, questioning whether exporting appendPQExpBufferVA from libpq was justified or StringInfo would be preferable #4.
- Tom Lane supported exit-on-allocation-failure behavior for application code but warned that widespread conversions would complicate backpatching, suggesting identical APIs with different failure behavior; his comments addressed the buffer-policy question rather than reviewing the test-driver patch #5.
- The test-driver patch from #4 subsequently landed unchanged as e70abd67c3, while the supplied thread leaves the pg_upgrade patch and broader buffer-policy questions unresolved.