)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":7847,"name":"Alistair Coles","email":"alistairncoles@gmail.com","username":"acoles"},"change_message_id":"82154b8ee4c3f07d29edf488199ae7f29aa24684","unresolved":true,"context_lines":[{"line_number":12,"context_line":"a no-jitter DELETE or POST, because \"\u003cfloat\u003e\" sorts before"},{"line_number":13,"context_line":"\"\u003cfloat\u003e_2jjj...\", so the PUT wins."},{"line_number":14,"context_line":""},{"line_number":15,"context_line":"This inverts the previous same-deca-microsecond tie-break. Before"},{"line_number":16,"context_line":"jitter a PUT and a DELETE/POST in the same deca-microsecond had equal"},{"line_number":17,"context_line":"timestamps, so on disk a tombstone trumped the data and in a container"},{"line_number":18,"context_line":"the surviving deleted flag depended on the order rows were merged. The"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"572cedaa_f93c64aa","line":15,"updated":"2026-06-29 09:33:02.000000000","message":"\"previous\" here is referring to \"before v2 timestamps jitter\", not \"before this patch\"\n\nThis subtlety is moot if this patch is squashed into the (grand)parent v2 timestamps patch.","commit_id":"04203206d68352c03a4cfcae199291e1e8a0e247"},{"author":{"_account_id":7847,"name":"Alistair Coles","email":"alistairncoles@gmail.com","username":"acoles"},"change_message_id":"82154b8ee4c3f07d29edf488199ae7f29aa24684","unresolved":true,"context_lines":[{"line_number":14,"context_line":""},{"line_number":15,"context_line":"This inverts the previous same-deca-microsecond tie-break. Before"},{"line_number":16,"context_line":"jitter a PUT and a DELETE/POST in the same deca-microsecond had equal"},{"line_number":17,"context_line":"timestamps, so on disk a tombstone trumped the data and in a container"},{"line_number":18,"context_line":"the surviving deleted flag depended on the order rows were merged. The"},{"line_number":19,"context_line":"change is easy to miss and was untested."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Add a diskfile test that a jittered .data trumps a same-deca-microsecond"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"3a5a4d82_90e325fc","line":18,"range":{"start_line":17,"start_character":12,"end_line":18,"end_character":65},"updated":"2026-06-29 09:33:02.000000000","message":"this sounds like a *bug* ? could we end up with the container row being inconsistent with on-disk state when PUT and DELETE timestamps are equal (with or without jitter)? If so, we should fix that regardless of jitter i.e. fix on master","commit_id":"04203206d68352c03a4cfcae199291e1e8a0e247"}]}