Review main Merged

89293[BUGFIX] Honor parameter types in writable Connection methods

Stefan Bürk CR +2 V +2 Hard +385 −22 updated 07/23/26, 07:50

Commit message

With #99062 native json field support has been introduced
along with automatic and forceful to-database value transformation
when using the following methods:

* `Connection::insert()`
* `Connection::update()`
* `Connection::bulkInsert()`

The intent of this change was to automatically map native PHP data
structures like `array` to JSON fields, but also introduced other
convenience mappings, like supporting automatic `\DateTime`
transformation to database values.

In other words, our raw connection API became a mini ORM, that
was automatically triggered based on database introspection.
Note: upstream doctrine `Connection` API supports this transformation
as well, but only if the parameter type (that does the actually
transformation) has been passed explicitly, while we autodetected
the type and applied it regardless of possible user input.

This forceful automatism broke consumers that already used JSON
database types, because `json_encode` is not idempotent and a
string that was already json encoded became double json encoded,
as it was passed a second time through `json_encode`.

Without taking this automatism away, we now restrict it to be
optional and no longer forceful, to only apply if no explicit doctrine
type parameter is given.

If a consumer of the `Connection` methods supplies a parameter type, it
is now honored and therefore allows to perform raw insertion, by using
doctrine binding types (`\Doctrine\DBAL\ParameterType::*`) instead of
doctrine mapping types (`\Doctrine\DBAL\Type\*`).

This improved handling is now used by Extbase (which does a to-database
transformation on its own) to prevent double encoding of already
prepared database field values:

The Extbase `Typo3DbBackend` uses the cached SchemaInformation
to pass the binding types `ParameterTypes` for all fields,
to ensure that data is inserted as-is, while preserving support
for all databases that require LOB handling for binary fields.

In the future, we may remove the Extbase internal handling
and use the `Doctrine DBAL` type conversion directly and
reduce a maintenance level, which
needs a little bit more of preparation work and brain to
achieve that.

Tests with Extbase usage and persisting models with loaded
data containing json encoded strings were missing and lead
to the issues and the reason to put additional tests with
this change into place, to have at least these cases covered.

Review

  • Code review CR +2 Done
    • Stefan Bürk +2
    • Benjamin Franzke +1
    • Garvin Hicking +1
    • Simon Schaufelberger +1
  • Verified V +2 Done
    • Stefan Bürk +2
    • Benjamin Franzke +1
    • Oliver Klee +1
    • Garvin Hicking +1
    • Simon Schaufelberger +1
    • core-ci +1 CI

Issues 3

Files 23

Show 3 more files

Review log 80119

Show 5994 earlier messages
  1. Stefan Bürk uploaded patch set 1 05/04/25, 19:36

  2. core-ci voted on patch set 1 V -1 05/04/25, 19:42 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/78493

  3. Stefan Bürk uploaded patch set 2 V -1 05/04/25, 19:44

  4. core-ci voted on patch set 2 V -1 05/04/25, 19:50 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/78494

  5. Stefan Bürk uploaded patch set 3 V -1 05/04/25, 19:53

  6. core-ci voted on patch set 3 V +1 05/04/25, 19:59 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/78495

  7. Stefan Bürk updated the commit message in patch set 4 V +1 05/04/25, 21:50

  8. core-ci voted on patch set 4 V +1 05/04/25, 21:56 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/78498

  9. Benni Mack voted on patch set 4 CR +1 V +1 05/04/25, 22:15

  10. Anja Leichsenring voted on patch set 4 CR +1 V +1 05/05/25, 05:34

  11. Benjamin Franzke commented on patch set 4 CR -1 05/05/25, 05:43

    v12 is in priority bugfix mode. Can you elaborate how this qualifies as a priority BUG? Arguments against: * It never worked since the introduction of type=json * There is a existing solution by using a TEXT column, which most will have used for this usecase * I would not even consider this a valid usecase anyway, as we should rather work on supporting type=json in extbase. So if we do sth. about this, then I propose to go v13 only.

    Connection.php:437

    I do not like this heuristic and the workaround to hotfix this in EXT:core instead of extbase at all. JSON is not just encoded arrays or objects, but also encoded simple types can be be encoded into all sorts of strings, where the resulting string would be double encoded as well: ``` php > var_dump(json_encode(null)); string(4) "null" php > var_dump(json_encode(1)); string(1) "1" php > var_dump(json_encode('foo')); string(5) ""foo"" ``` Even if extbase does not know about type=json yet, it has a column map, could parse it and for type=text in calls to our databse layer.. Whatever we do, I think this hack misses a plan for when to phase this out, since this extbase-oriented hack effectively becomes API for all our doctrine users and is applied there as well. That means: Devs may dump JSON strings (e.g. for API) as-is and once we drop this hack, it could break these queries :\
  12. Stefan Bürk abandoned the change 05/05/25, 07:11

  13. Stefan Bürk commented on patch set 4 05/05/25, 07:21

    Connection.php:437 Reply

    Abonding the change. going create the v14 breaking removal of the feature somewhere the next days.
  14. Stefan Bürk restored the change 10/08/25, 10:13

  15. Stefan Bürk marked it work in progress 10/08/25, 10:13

  16. Stefan Bürk uploaded patch set 5 CR +1 V +1 10/08/25, 11:14

  17. Stefan Bürk commented on patch set 5 10/08/25, 11:19

    Reply

    Ack. And it should also be possible to use composer patches to backport the change to v12 manually if required, special after reworking the change.

    Connection.php:437 Reply

    Closing this comment chain now. The remarks were valid and got me to rethink the whole approach. The new approach does not use such heurestic determinations anymore and change needs to be reviewed new completly. The pointed out code does not exists anymore, so dropping it for now.
  18. core-ci voted on patch set 5 V +1 10/08/25, 11:21 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86225

  19. Stefan Bürk uploaded patch set 6 V +1 10/08/25, 16:18

  20. core-ci voted on patch set 6 V +1 10/08/25, 16:26 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86242

  21. Stefan Bürk uploaded patch set 7 V +1 10/08/25, 16:30

  22. core-ci voted on patch set 7 V +1 10/08/25, 16:38 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86244

  23. Stefan Bürk uploaded patch set 8 V +1 10/08/25, 17:58

  24. core-ci voted on patch set 8 V +1 10/08/25, 18:08 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86246

  25. Stefan Bürk rebased onto patch set 9 V +1 10/08/25, 18:40

  26. core-ci voted on patch set 9 V +1 10/08/25, 18:54 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86254

  27. Benjamin Franzke commented on patch set 9 10/10/25, 11:17

    Typo3DbBackend.php:93

    Hm, this would prevent other fields like native BOOLEAN fields to be encoded for the the postgresql platform. Extbase DataMapper encodes these as 1/0 (which is good for mysql/sqlite), which would then be incompatible with postgres, if the doctrine type does not convert. Note: I didn't test this, this is just a theoretical analysis, maybe I'm wrong… I guess we actually need to: * Introspect the keys of `$fieldValues` * check if it has a *json* dbtype via our new SchemaInformation cache * check if the value is a *string* * Signal to the connection layer that specific fields are not to be converted to database format (technically that could be a list of field names: `$preprocessedDatabaseValueFieldNames: list<string>`)
  28. Stefan Bürk rebased onto patch set 10 V +1 10/10/25, 17:28

  29. core-ci voted on patch set 10 V +1 10/10/25, 17:40 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86429

  30. Stefan Bürk rebased onto patch set 11 V +1 10/11/25, 16:12

  31. core-ci voted on patch set 11 V +1 10/11/25, 16:21 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86469

  32. Stefan Bürk commented on patch set 11 10/13/25, 08:05

    Typo3DbBackend.php:93 Reply

    Native BOOLEAN fields are not supported by our TCA/ext_tables.sql/PSR-14 event API, as BOOLEAN is not parsed and mapped to Doctrine DBAL `BooleanType`, which means that in that case it will always try to change the field (Database Analyzer). In case for external databases, completly managed outside and connection retrieved directly this may be correct - but are not usable with extbase, which requires that table names are defined as table mapping and ending in the Database Analyzer again (I guess, not tested). What you describe was not possible before v12 anyway and was a side-effect while introducing native JSON support and the automatic converting of values in `Connection` methods `insert()`, `update()` and `bulkInsert()` - and used in DataHandler and Extbase, including the automatic json encoding for database fields of type JSON (or TextField created as fallback for defined as native json field in ext_tables.sql/PSR-14-Event/TCA type=json but platform is used not supporting native json field). This change and also the followup https://review.typo3.org/c/Packages/TYPO3.CMS/+/90957 are ment as kind of (soft) revert of this feature which was not a good thought choise back then by myself, leaving conversion handling to highler level (DataHandler/Extbase) itself and supporting API, and making the lowlevel connecection free and in the hand of developers. In case of boolean fields, that means that projects should handle this on their own in case they are using specific, vendor based database support leaving TYPO3 API out and trying to use "some" api's again. If that is really an issue, we should fix / handle boolean conversion in Extbase ORM than directly instead of blindly converting it to integer `0/1` in the long term. > I guess we actually need to: Sorry, what you suggest here now does not work and is not really usable, beside that it requires kind of heurestic detection you already emphasized that you dislike a heurestic detection approach, see https://review.typo3.org/c/Packages/TYPO3.CMS/+/89293/comment/78d746e4_7651f2f8/ To tell if a string is a "json encoded value" or not requires a heurestic implementation, checking for quite a lot of stuff .. and after your abovementioned comment about disliking the heurestic I must say that I also don't feel that this would be a good approach. Converting scalar values may be a good thing to keep, which would mean that we should use the new 4th argument(flag) only liiting special type convertions (JsonType, DateTimeType, DateTimeImuttableType, ...) - or only for the JsonType. Note, that even if the issue has been reported through extbase thas also hits if a row is retrieved containing a json value (string), not decoding it and using `Connection::insert()`/`Connection::update()` with that value, ending up in double encoding the value like revealed by the extbase usage. Something I consider a riskier and the reason why I'm trying to restore from that by making the option in v14 false and a optional/explicitly to use thing on the lowlevel connection - if not deprecating and remove it in v15 again (still think keeping it optional may be okay). I would say that native postgres boolean fields are really an edge case, if used by anyone and I would still go with this change to start the revert and prevent the issue.
  33. Stefan Bürk uploaded patch set 12 V +1 10/13/25, 10:04

  34. core-ci voted on patch set 12 V +1 10/13/25, 10:14 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86526

  35. Stefan Bürk rebased onto patch set 13 V +1 10/16/25, 07:45

  36. core-ci voted on patch set 13 V +1 10/16/25, 08:02 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/86789

  37. Stefan Bürk rebased onto patch set 14 V +1 11/10/25, 12:25

  38. core-ci voted on patch set 14 V +1 11/10/25, 12:40 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/88841

  39. Stefan Bürk uploaded patch set 15 V +1 11/10/25, 15:58

  40. core-ci voted on patch set 15 V -1 11/10/25, 16:16 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/88873

  41. Benjamin Franzke commented on patch set 15 11/10/25, 16:45

    Connection.php:425

    We can omit the fallback type, because it is the default doctrine parameter type and we shouldn't explicitly set a default. ```suggestion if ($tableInfo->hasColumnInfo($key)) { $types[$key] = $tableInfo->getColumnInfo($key)->getType(); } ```
  42. Stefan Bürk commented on patch set 16 V -1 11/10/25, 17:45

    Published edit on patch set 15.
  43. Stefan Bürk commented on patch set 16 11/10/25, 17:46

    Connection.php:425 Reply

    Fix applied.
  44. Stefan Bürk uploaded patch set 17 11/10/25, 17:57

  45. core-ci voted on patch set 16 V -1 11/10/25, 18:05 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/88889

  46. core-ci voted on patch set 17 V -1 11/10/25, 18:13 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/88890

  47. Stefan Bürk uploaded patch set 18 V -1 11/10/25, 18:53

  48. core-ci voted on patch set 18 V +1 11/10/25, 19:01 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/88896

  49. Garvin Hicking commented on patch set 18 11/10/25, 22:17

    Typo3DbBackend.php:664

    ```suggestion ``` We could remove this redundant type hint due to the same type declaration
  50. Stefan Bürk uploaded patch set 19 V +1 11/11/25, 09:13

  51. Stefan Bürk uploaded patch set 20 11/11/25, 09:14

  52. Stefan Bürk commented on patch set 20 11/11/25, 09:15

    Typo3DbBackend.php:664 Reply

    Done
  53. core-ci voted on patch set 19 V +1 11/11/25, 09:24 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/88962

  54. core-ci voted on patch set 20 V +1 11/11/25, 09:27 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/88963

  55. Benjamin Franzke commented on patch set 20 11/11/25, 11:13

    Typo3DbBackend.php:93 Reply

    Done
  56. Stefan Bürk rebased onto patch set 21 V +1 11/11/25, 17:58

  57. core-ci voted on patch set 21 V +1 11/11/25, 18:07 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/89047

  58. Benjamin Franzke uploaded patch set 22 V +1 11/12/25, 12:21

  59. core-ci voted on patch set 22 V -1 11/12/25, 12:29 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/89308

  60. Benjamin Franzke uploaded patch set 23 V -1 11/12/25, 16:29

  61. Benjamin Franzke commented on patch set 23 11/12/25, 16:29

    tx_testjsonfields_domain_model_example.php:38

    BTW, this field (native json, text in form engine) double encodes in FormEngine. When I enter ``` { "foo": "bar" } ``` It is changed to the following value after save: ``` "{\r\n \"foo\": \"bar\"\r\n}" ```
  62. Benjamin Franzke commented on patch set 23 11/12/25, 16:33

    tx_testjsonfields_domain_model_example.php:38 Reply

    For the record: Also happens without this patch.
  63. Stefan Bürk marked it work in progress 11/12/25, 16:38

  64. core-ci voted on patch set 23 V +1 11/12/25, 16:39 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/89369

  65. Stefan Bürk uploaded patch set 24 V +1 11/13/25, 17:36

  66. Stefan Bürk commented on patch set 24 11/13/25, 17:37

    tx_testjsonfields_domain_model_example.php:38 Reply

    Added a comment to the TCA configuration that extbase is the test target and not form engine. Okay for you @ben@bnf.dev ?
  67. core-ci voted on patch set 24 V +1 11/13/25, 17:44 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/89557

  68. Stefan Bürk uploaded patch set 25 V +1 05/27/26, 13:32

  69. core-ci voted on patch set 25 V -1 05/27/26, 13:44 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/102483

  70. Stefan Bürk uploaded patch set 26 V -1 05/28/26, 07:55

  71. core-ci voted on patch set 26 V -1 05/28/26, 08:06 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/102512

  72. Stefan Bürk uploaded patch set 27 V -1 05/28/26, 08:32

  73. Stefan Bürk uploaded patch set 28 05/28/26, 08:33

  74. core-ci voted on patch set 27 V -1 05/28/26, 08:50 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/102518

  75. core-ci voted on patch set 28 V -1 05/28/26, 09:10 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/102519

  76. Stefan Bürk uploaded patch set 29 V -1 05/28/26, 09:13

  77. Stefan Bürk uploaded patch set 30 05/28/26, 09:19

  78. core-ci voted on patch set 29 V -1 05/28/26, 09:22 Core CI is not happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/102526

  79. core-ci voted on patch set 30 V +1 05/28/26, 09:29 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/102528

  80. Stefan Bürk rebased onto patch set 31 V +1 05/28/26, 14:01

  81. Stefan Bürk commented on patch set 31 05/28/26, 14:05

    This change is ready for review.

    tx_testjsonfields_domain_model_example.php:38 Reply

    Other issue - or to be precise DataHandler related - started working in it with a dedicated change and similar but unrelated to extbase issue. Going to handle that in a dedicated change/chain.
  82. core-ci voted on patch set 31 V +1 05/28/26, 14:39 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/102547

  83. Oliver Klee voted on patch set 31 V +1 05/29/26, 06:46

  84. Stefan Bürk rebased onto patch set 32 V +1 06/02/26, 12:11

  85. core-ci voted on patch set 32 V +1 06/02/26, 12:18 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/102828

  86. Oliver Klee voted on patch set 32 V +1 06/02/26, 14:29

  87. Stefan Bürk uploaded patch set 33 V +1 06/12/26, 07:56

  88. core-ci voted on patch set 33 V +1 06/12/26, 08:05 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/103433

  89. Garvin Hicking updated the commit message in patch set 34 V +1 06/12/26, 08:30

  90. Stefan Bürk updated the commit message in patch set 35 06/12/26, 08:37

  91. core-ci voted on patch set 34 V +1 06/12/26, 08:38 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/103435

  92. core-ci voted on patch set 35 V +1 06/12/26, 08:45 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/103436

  93. Oliver Klee voted on patch set 35 V +1 06/12/26, 09:32

  94. Stefan Froemken commented on patch set 35 06/12/26, 10:36

    database
    users?
  1. Stefan Froemken commented on patch set 35 06/12/26, 10:50

    Connection.php:432

    If a field does not exist in Schema, no kind of ParameterType is needed.

    composer.json:5

    Yes, it "is" allowed here, but I prever to set version in extra.typo3/cms. https://getcomposer.org/doc/04-schema.md#version "should be omitted"

    AddTest.php:322

    I don't understand that. Why testing with json_encoded data. Isn't this the job of the framework itself?
  2. Garvin Hicking updated the commit message in patch set 36 V +1 06/12/26, 12:16

  3. Garvin Hicking commented on patch set 36 06/12/26, 12:20

    Reply

    Done

    Reply

    Done

    Connection.php:432 Reply

    Doctrine does this fallback IIRC.

    composer.json:5 Reply

    All other fixture extensions specify this version like that. This is not really meant as an "actual" extension version. We could open a distinct issue/discussion about this, but then should change it everywhere the same, so this is beyond the scope here.

    AddTest.php:322 Reply

    This is a test to ensure no double JSON encoding takes place even when inserting JSON encoded data into a field marked as JSON, AFAICS.
  4. Garvin Hicking voted on patch set 36 CR +1 V +1 06/12/26, 12:22

  5. core-ci voted on patch set 36 V +1 06/12/26, 12:23 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/103447

  6. Oliver Klee voted on patch set 36 V +1 06/12/26, 14:02

  7. Benjamin Franzke commented on patch set 36 06/16/26, 10:00

    Hey, in my opinion we should avoid such an impactful change in priority bugfix mode. We decided against that already in https://review.typo3.org/c/Packages/TYPO3.CMS/+/89293/comment/231f86d5_c3de652b/ (by that time it was v12, but not it is v13, but I think the reasoning remains the same). What do you think, maybe some projects which really need this in v13?

    Connection.php:434

    This graceful fallback (instead of prior enforcement) is much better now, thanks!

    Typo3DbBackend.php:678

    I'm not sure this is right (didn't test yet): When we talked about this (yes, I know, back than in october 2025), i think we concluded, that we'd need to send a json encoded string as `Doctrine\DBAL\Types\TextType` instead of `Doctrine\DBAL\Types\JsonType` in order to avoid double json encoding. If we now introspect and detect `JsonType` based on the schema, we should detect `JsonType` and therefore still produce the double encoding. I know this patch adds test and I may be totally wrong, but I want to mention it anyway, since I basically do not understand this part right now.
  8. Stefan Bürk commented on patch set 36 06/18/26, 10:03

    Reply

    Having mentionings with v12 and v13 in public channels with these issues using extbase yes. But I'm not hard relaying on it. I just want to get that in and if it is main only okay. Meanwhile it's a common pattern to have a collection of important core fixes manually backported in project and common practise in agencies. Not fighting for it. If the change / fix is acceptable for any version I'm going with any reducing on which version/branch it lands.

    Typo3DbBackend.php:678 Reply

    Extbase has no (native) json support on models. That means that JSON (JsonType) ends as `string` properites in the model. Detecting from the database schema and using `JsonType` (in https://review.typo3.org/c/Packages/TYPO3.CMS/+/89293/36/typo3/sysext/core/Classes/Database/Connection.php#435) this leads to double encoding. We refactored the other place to have the ability to "override" it. Here the JsonType is detected from the database schema but instead of basing JsonType using the convert to databsae method on the type to encode the value (double encoding) we pass here the `binding type)` which is `ParameterType::STRING`. `ParameterType::STRING` does not handle anyconvertion and let the string simple be passed through. Simple said: With the code here wie replace `JsonType` with `ParameterType::string` and dismantle the json encoding handed down below in doctgrine. (Override of the type handling). Reason: Extbase is in charge of handle json encode/decoding and not the connection/dbal layer. Does this help to remember or clarify this here ? Let me know if not.
  9. Stefan Bürk updated the commit message in patch set 37 CR +1 V +1 07/18/26, 00:20

  10. Stefan Bürk rebased onto patch set 38 07/18/26, 00:20

  11. core-ci voted on patch set 37 V +1 07/18/26, 00:28 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/104918

  12. core-ci voted on patch set 38 V +1 07/18/26, 00:29 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/104919

  13. Stefan Bürk commented on patch set 38 07/18/26, 00:30

    Reply

    Removed `13.4`.
  14. Stefan Bürk commented on patch set 38 CR -1 07/20/26, 20:00

    Typo3DbBackend.php:678 Reply

    Okay - you are possible right. -1 and checking again.
  15. Stefan Bürk commented on patch set 38 CR 0 07/21/26, 12:31

    Typo3DbBackend.php:678 Reply

    Okay, that's okay and correct here to use the binding type. The point is, that `extbase` is in charge of dealing with conversion, like DateTime handling and also json encode/decode (albeit currently not doing so); currently with plain extbase json can only be retrieved as `string` (in a property) and would lead to the double json encoding due to the "database schema based lookup" in `Connection` (changed with this change). The overall strategy we discussed back than was: * `Connection` only set doctrine types (not binding types) if not provided manually as type specification (would use JsonType for a json field) and doing the conversion. * Here in Typo3DbBackend (Extbase) we do the same lookup and use the binding type and prevent the doctrine value to dbvalue conversion (json_encode() for json fields) - because that is the job of Extbase (in the future) and be the encoded value at this point already. Maybe let's have a call to talk (clarify) that (again) - end comment the outcome together here afterwards.
  16. Benjamin Franzke commented on patch set 38 CR +1 07/22/26, 12:50

    Typo3DbBackend.php:678 Reply

    Alright, thanks for clarification and the talk. To summarise: My misunderstanding was that I assumed that `getBindingType()` would return `Doctrine\DBAL\Types\JsonType`, while it actually returned `Doctrine\DBAL\ParameterType::STRING`. So, `getType()` indeed returns `JsonType`, the chained call resolved to `getBindingType()` ensures that we end up with a plain parameter type, that does not cause implicit transforms later on (that's what we agreed back than and what Stefans code achieves 👌). Resolved from my side, thanks for you patience and sorry that this required another loop. We agreed to add a specific comment, in order to help others, when stumpling over the same thing, like I did. Thanks!
  17. Stefan Bürk uploaded patch set 39 CR +1 V +1 07/22/26, 13:21

  18. core-ci voted on patch set 39 V +1 07/22/26, 13:33 Core CI is happy: https://git.typo3.org/typo3/CI/cms/-/pipelines/105184

  19. Benjamin Franzke voted on patch set 39 CR +1 07/22/26, 13:36

  20. Benjamin Franzke voted on patch set 39 V +1 07/22/26, 13:38

  21. Simon Schaufelberger voted on patch set 39 CR +1 V +1 07/22/26, 18:23

  22. Garvin Hicking voted on patch set 39 CR +1 V +1 07/22/26, 20:35

  23. Oliver Klee voted on patch set 39 V +1 07/23/26, 06:58

  24. Stefan Bürk voted on patch set 39 CR +2 V +2 07/23/26, 07:09

  25. Stefan Bürk merged patch set 40 07/23/26, 07:50