Review main Ready for review Quiet Breaking

96653[!!!][TASK] Restrict TCA overrides for types and special values

Daniel Windloff CR 0 V -1 Expert +1116 −76 updated 10/26/26, 04:30

Commit message

a) Respect column config type values
Remove all column > config > type values from
- overrideChildTca
- columnsOverrides

Respect all column > config > type values set in
any Configuration/TCA file or subsequent core functionality
like TcaMigration, TcaEnrichment or TcaPreparation.

b) Deny changes for specific TCA column config
Set a fixed list of TCA configuration to specific values
and remove any possible type override of these values
in columnsOverrides.
These values are already enforced in DataHandler by
special handling. Enforcing them, makes it understandable
why the processing acts this way checking the resulting TCA.
Remove that configuration from DataHandler.

Test this change

Run a command in your clone of the TYPO3 Core. Then try what the change does and vote on Verified in Gerrit. How to review a patch

Check out patch set 7
git fetch https://review.typo3.org/Packages/TYPO3.CMS refs/changes/53/96653/7 && git checkout FETCH_HEAD
Cherry-pick onto main
git fetch https://review.typo3.org/Packages/TYPO3.CMS refs/changes/53/96653/7 && git cherry-pick FETCH_HEAD

Review

  • Code review CR 0 Needs a vote
  • Verified V -1 Blocked
    • core-ci -1 CI

Merge conflicts 38

Issues 1

Files 32

Show 12 more files

Review log 1219

  1. Daniel Windloff uploaded patch set 1 10/13/26, 04:10

  2. core-ci voted on patch set 1 V -1 10/13/26, 04:43 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/110961

  3. Daniel Windloff uploaded patch set 2 V -1 10/13/26, 05:10

  4. core-ci voted on patch set 2 V +1 10/13/26, 05:20 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/110970

  5. Daniel Windloff uploaded patch set 3 V +1 10/16/26, 04:48

  6. core-ci voted on patch set 3 V -1 10/16/26, 04:59 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/111309

  7. Daniel Windloff rebased onto patch set 4 V -1 10/16/26, 11:27

  8. core-ci voted on patch set 4 V +1 10/16/26, 11:35 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/111351

  9. Daniel Windloff rebased onto patch set 5 V +1 10/23/26, 15:17

  10. core-ci voted on patch set 5 V -1 10/23/26, 16:25 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/112342

  11. Daniel Windloff rebased onto patch set 6 V -1 10/24/26, 02:50

  12. core-ci voted on patch set 6 V +1 10/24/26, 03:01 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/112377

  13. Garvin Hicking commented on patch set 6 10/24/26, 08:02

    I think there are scenarios that result in DB schema changes which we should block (like turning selects into inputs). But there are several changes I see useful for integrators to fit rules, and which I do see commonly used in projects: - turn text to input - turn input to text - turn text to select (for restricted presets) - turn anything into passthrough Since you can "own" a type change with ext_tables.sql thats also interesting for overriding third party code with your own intent.
  14. Daniel Windloff commented on patch set 6 10/24/26, 10:12

    Reply

    I would not question that there are several valid use cases for overriding a `type`. The approach here is from a different perspective: For the refactoring of the `DataHandler`, specifically to eliminate core hooks or to integrate hard-coded logic (such as page permissions, language diff sources, etc. ) into the standard `process_datamap` workflow, we need a clear definition of when TCA overrides are permissible (an which kind) and when they are not. Many modifications that are currently allowed would simply lead to errors. For instance, changing the recordtype of `tt_content` to `input` might be technically feasible at the database level, but it would fail elsewhere. The same applies to almost all management fields and likely to fields like `password`, `slug`, etc.—which should not be redefined as `input`. In my view, allowing or prohibiting changes at the type level is also problematic, as it cannot represent the relationships between columns. The current restriction was an attempt to establish a rule, admittedly imperfect, but easy to follow, for this purpose. I have outlined an earlier idea with the same goal here: https://forge.typo3.org/issues/110720 . In my view, the system cannot function without a reasonable, established baseline for configuration. Hardcode this in DataHandler hooks only leads to inconsistent behavior between different parts of the core like FormEngine, DefaultSchema or Extbase. That is why I suggest changing TCA directly early. This does not necessarily mean that the proposal presented here is the only way to achieve that. The current configuration overrides allows anything — including the possibility of breaking the system. I don't think that should work ;)
  15. Garvin Hicking commented on patch set 6 10/24/26, 12:12

    Reply

    Short on time atm, but: could we recommend to use a different rendertype? Eg keep base type and adjust the display from input=>text and vice versa? We'd "just" need to ensure that rendertypes are interchangeable for every type? At least for my use cases that could work (haven't tried) and we might have smaller "annoyance impact" from others. Usually I think people just want to change display/represantion but not persistence....?
  16. Daniel Windloff commented on patch set 6 10/24/26, 18:53

    Reply Unresolved

    Exactly. Strictly speaking, as far as I’m aware, `renderTypes` were actually introduced for this very reason - adding different representation of the same value. Changing input to text could be still difficult, as the persistance ins't the same (e.g. rte transformations for example).
  17. Daniel Windloff uploaded patch set 7 V +1 10/26/26, 04:17

  18. Daniel Windloff wrote 10/26/26, 04:18

    Topic set to DataHandler
  19. core-ci voted on patch set 7 V -1 10/26/26, 04:30 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/112554