93025[TASK] Use BIGINT storage for sys_history, sys_log uid column
Commit message
Review
-
Code review CR -1
- Oliver Klee +1
- Benni Mack -1
-
Verified V +1
- Oliver Klee +1
- core-ci +1 CI
Issues 1
Files 2
Review log 911
-
Garvin Hicking uploaded patch set 1 02/27/26, 15:52
-
core-ci voted on patch set 1 V -1 02/27/26, 16:03 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/96977
-
Garvin Hicking commented on patch set 2
V -102/27/26, 16:13Published edit on patch set 1. -
Oliver Klee voted on patch set 2 CR +1 02/27/26, 16:19
-
core-ci voted on patch set 2 V +1 02/27/26, 16:23 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/96979
-
Oliver Klee voted on patch set 2 V +1 02/27/26, 16:45
-
Benni Mack voted on patch set 2 CR -1 03/06/26, 12:06
-
Stefan Bürk commented on patch set 2 03/06/26, 13:42
I'm not realy confident to backport this to 13.4 and even more to 12.4. First, instances can do that on their own with a sitepackage or extension in the project so there is no "real" need to do so. It's a "db" change on tables which may have more entires and DDL on a patchlevel update may become extensive due to that operation. Second, albeit considering it possible as low to no risk we don't have real live experiences with this and should abstain to introcue this to older versions. For main I'm fine with so agencies updating to 14.x/LTS have a longer path and can test this taking time on updates. Patchlevel releases however are more ment to be handled quicker. `sys_history` reaching already the INT limit would be really a boomer, `sys_log` and messenger it could be possible. I guess we should investigate for TYPO3 v15 to have non autoincrement for such tables but still a unique PK (using uuid for example) to avoid these kind of things, but that requires more work and investigation throughout all API's. -
Stefan Bürk commented on patch set 2 03/06/26, 13:44
Reply
Q: Not verified but I have in mind that the BIGINT max is not representable as PHP int value and would use "string" .. would that not break our api's in case ids are casted / passed to methods with string type declarations ? -
Garvin Hicking abandoned the change 03/09/26, 07:31
BIGINT makes PHP API go brrrr -
Garvin Hicking commented on patch set 2 03/11/26, 21:16
Reply Unresolved
Totally right, I forgot about this. I guess we'll have to go the road of introducing UUID then, with a breaking change. Abandoning this one.