model-writtenrun 6353 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- The documentation report in #1 described HDBC-postgresql failing to parse a historical Mexico City timestamp with a seconds-resolution UTC offset and requested documentation of that possible output.
- Tom Lane submitted #2 to document fractional-minute offsets in the main date/time documentation, adding input examples and explaining numeric offset formats for ISO and other output styles.
- Thomas Munro explained in #3 that the example came from IANA's historical local mean time data and noted its disclaimers about the meaningfulness of those offsets; this explanation did not review the attached patch.
- Tom Lane noted in #4 that his earlier proposal to remove local mean time from the upstream timezone data had attracted no support.
- Tom Lane committed #2 as ab2e19987f, with light wording edits and relocation of the explanatory paragraph.
- After the documentation change landed, Thomas Munro argued in #5 that extending standard time backward or rejecting undefined conversions would be preferable to questionable historical local mean time values; the thread ended without a further patch.