model-writtenrun 6025 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Robert Haas reported that a replication connection opened before standby promotion could copy an uninitialized ThisTimeLineID during slot creation, although subsequent code appeared to overwrite it before use; even substituting a random timeline passed check-world #1.
- Andres Freund proposed an explicit invalid-timeline constant and assertions, while Alvaro Herrera favored regular xlogreader use for physical replication and stronger tests #2 #3.
- Robert Haas argued that the initialization and its explanatory comment were redundant because the bogus values were overwritten before use, while Alvaro Herrera maintained that test coverage also needed improvement #4 #5.
- Michael Paquier noted that walsender already treated zero as an invalid timeline and supported assertions; Alvaro Herrera reiterated the preference for a named constant #6 #7.
- Robert Haas submitted a patch deleting the redundant initialization after tracing how logical_read_xlog_page initializes the same globals before WALRead; he separately proposed replacing direct ThisTimeLineID access with explicit arguments or an accessor, leaving that broader redesign outside this patch #8.
- Robert Haas committed the attached deletion unchanged as 902a2c2800 and announced completion #9; the commit explicitly avoided backpatching because there was no functional defect, and the broader timeline-handling proposals remained unimplemented in this thread.