Review main Abandoned

63354[FEATURE] Introduce nested content grids

Claus Due CR 0 V 0 Expert +1696 −12 updated 04/27/21, 18:39

Commit message

Adds support for defining "content grids" to use with specific
content types. Selecting a content grid then becomes possible
in those content types, and the grid layout is rendered as a
sub-grid in the page , with full support for drag and drop,
localization, etc.

Review

  • Code review CR 0
  • Verified V 0

Issues 1

Files 25

Show 5 more files

Review log 126165

Show 109140 earlier messages
  1. Claus Due uploaded patch set 1 02/21/20, 18:34

  2. TYPO3com voted on patch set 1 V -1 02/21/20, 18:41 Completed build in 37s on Fri, 21 Feb, 06:41 PM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-425

  3. Claus Due uploaded patch set 2 02/21/20, 18:42

  4. TYPO3com voted on patch set 2 V +1 02/21/20, 19:02 Completed build in 19m 30s on Fri, 21 Feb, 07:02 PM Test Summary: 71476 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-426

  5. Benni Mack commented on patch set 2 02/21/20, 19:52

    Aah, a basis to discuss! Thanks for pushing it. Shouldn't we use "Content Layouts" instead of "Page Backend Layouts" and have a separation here?
  6. Claus Due commented on patch set 2 02/21/20, 20:17

    > Patch Set 2: > > Aah, a basis to discuss! Thanks for pushing it. Shouldn't we use "Content Layouts" instead of "Page Backend Layouts" and have a separation here? I would aim for naming that clearly distinguishes between: * Content that is child content (inserted in a parent) * Content that is "root column level" (content w/o any parent) * Grid Layouts that are page-level * Grid Layouts that are content-level In the patch I went with naming: * "Sub content" when dealing explicitly with the content rendering FE output * Page Backend Layouts when referring to root-level backend layouts * (Nested) Content Grid when referring to a backend layout used in content as opposed to page * "Child content" when referring to "Sub content" from a code context, e.g. in DataHandler, to distinguish this from FE "Sub content". The name "Child content" could also be changed in code context to use just "Sub content" as it conveys the same meaning, if it isn't a problem that FE and code contexts use the same names. There are some terms I think we should avoid: * "Grid Elements" since this might be confused with the same-named extension * "Grid Structure" since this is ambiguous and refers to the structure inside a backend_layout, whether used with page or content * "Content Layouts" since this can be confused with the "layout" field of tt_content (though this is rarely used these days) I do need to check for consistency of wording but my idea was to use "sub content" when speaking of FE rendering, "page backend layout" when speaking of root-level backend layouts, and "content grid" to refer to backend layouts used with content, to distinguish this from "backend layouts". The terminology would then be: * "Sub content": any content element placed within another content element * "Page backend layout": a backend_layout used at page level * "Content grid": a backend_layout used to define a grid within a content element These three terms would be nicely separated and hard to confuse with one another, and risks no confusion with third party extensions like gridelements. The drawback is perhaps that "Content grid" is slightly ambiguous in that it might be taken to mean "a grid containing content" which can be confused with a page-level backend layout, but if we clearly refer to page-level backend layouts as "Page backend layout" this *should* minimise the ambiguity.
  7. Claus Due uploaded patch set 3 02/21/20, 22:19

  8. Claus Due wrote 02/21/20, 22:21

    Description set to "Added unit tests for about half of cascading DataHandler ops"
  9. TYPO3com voted on patch set 3 V -1 02/21/20, 23:42 Completed build in 11m 57s on Fri, 21 Feb, 11:42 PM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-442

  10. Riccardo De Contardi commented on patch set 3 02/21/20, 23:43

    > * "Content grid": a backend_layout used to define a grid within a content element What about "Content Backend Layout" so it would be consistent with "Page Backend Layout"?
  11. Claus Due uploaded patch set 4 02/22/20, 08:34

  12. Claus Due wrote 02/22/20, 08:34

    Description set to "Made new DataHandler code not run if feature is disabled"
  13. Kevin Appelt commented on patch set 3 02/22/20, 08:34

    If someone wants to test the feature without first having to setup the elements I have prepared a little extension here: https://github.com/kevin-appelt/bootstrapgrid I will add more features later today to make testing even easier.
  14. Claus Due commented on patch set 4 02/22/20, 08:43

    > Patch Set 3: > > > * "Content grid": a backend_layout used to define a grid within a content element > > What about "Content Backend Layout" so it would be consistent with "Page Backend Layout"? The problem I have with "Content Backend Layout" is that it would be unlikely to yield meaningful results in searches both external and on our docs site. The word "content" is extremely likely to also appear in docs and examples about "Page Backend Layouts" - the word "Grid" has the benefit that it is very infrequently used in the core and docs right now.
  15. TYPO3com voted on patch set 4 V -1 02/22/20, 08:44 Completed build in 31s on Sat, 22 Feb, 08:44 AM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-457

  16. Claus Due uploaded patch set 5 02/22/20, 08:45

  17. Claus Due wrote 02/22/20, 08:45

    Description set to "Fix CGL problem preventing tests from running"
  18. TYPO3com voted on patch set 5 V +1 02/22/20, 09:13 Completed build in 27m 54s on Sat, 22 Feb, 09:13 AM Test Summary: 71572 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-458

  19. Claus Due uploaded patch set 6 02/22/20, 10:25

  20. Claus Due wrote 02/22/20, 10:26

    Description set to "Fixed potentially missing language UID, added content_grid to copy-field-default-value, refactored a recursive method to prep for unit test"
  21. TYPO3com voted on patch set 6 V +1 02/22/20, 11:12 Completed build in 25m 59s on Sat, 22 Feb, 11:12 AM Test Summary: 71572 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-463

  22. Kay Strobach commented on patch set 6 02/22/20, 12:24

    some small things from reading the patch

    ColumnNumberUtility.php:31

    that maximum 100 column thingy should be named in the feature.rst

    DataHandler.php:3000

    is a break missing here?

    Feature-90444-NestedContentElementsGridElementsFeature.rst:4

    i would not call it "grid elements", as this misleads people thinking that "ext:gridelements" is now in the core. how about calling it "content grid" like done below

    backend_layout.php:49

    wouldn't it make sense to not use a displaycond here, but simply add the needed tca only if the feature is enabled? maybe this also adds too much complexity and also makes showitems more complex
  23. Kay Strobach commented on patch set 6 02/22/20, 12:29

    backend_layout.php:49 Reply

    Done
  24. Kay Strobach commented on patch set 6 02/22/20, 12:38

    DefaultConfigurationDescription.yaml:213

    Please also point out, that this is a technical preview, so that no one gets disappointed if we still find something we need to change
  25. Claus Due commented on patch set 6 02/22/20, 13:13

    ColumnNumberUtility.php:31 Reply Unresolved

    Good point, it absolutely should! I fact I am considering throwing an exception in BackendLayout if a colPos > 100 || colPos < 0 is detected, but the latter might cause problems for EXT:gridelements.

    DataHandler.php:3000 Reply Unresolved

    Correct, good catch! Fixed locally, will push with next PS.

    DefaultConfigurationDescription.yaml:213 Reply Unresolved

    Should I go further and indicate it as "experimental feature, may be subject to change"?

    Feature-90444-NestedContentElementsGridElementsFeature.rst:4 Reply Unresolved

    I will adapt naming across the patch and RST when we reach a concensus.

    backend_layout.php:49 Reply

    I considered this and we could do it like that; however, that implies creating a TCA/Overrides/backend_layout.php file. The inspiration to use a displayCond came from an existing case, \TYPO3\CMS\Redirects\UserFunctions\HitCountDisplayCondition
  26. Claus Due uploaded patch set 7 02/22/20, 16:33

  27. Claus Due commented on patch set 7 02/22/20, 16:33

    Changes in recent patch set: * Feature marked as "EXPERIMENTAL" in Feature Toggles description. * API method BackendLayoutView->getSubBackendLayoutForContentRecord marked as internal. * Wording in feature RST adjusted to avoid use of "grid elements". Content grids are now referred to only as "Content grids". * Documented pageTSconfig definition in feature RST. * Adjusted RST examples to void using "textmedia" for demonstration; theoretical third-party CType used instead. * ContentGridDataProvider interface added to split the new method that resolves content grids for a record, to a dedicated interface. * Removed support for defining content grids in the backend_layout table. Content grids are now exclusively possible to define as pageTSconfig, or deliver from a third party which implements a custom DataProvider which implements the ContentGridDataProvider interface. * Removed ability to resolve content grids from DefaultDataProvider and PageTsConfigDataProvider. * Added new ContentGridPageTsConfigDataProvider which reads definitions from mod.web_layout.ContentGrids to avoid mixing page-level backend layouts and content grids. * Removed "content_grid" field from TCA of all CTypes, making it so the only way to use the field would be to manually register it for use with a specific CType. * Subsequently removed the displayCond implementation that hides the field (since it is now fully opt-in). * Made SubContentDataProcessor not assign any variables if no content grid was selected in record being rendered. Credit for these changes go to Benji who spent many hours today helping me align and improve the feature. Fixes: * Fixed issue with usage of BackendLayout classes in frontend. * Added a missing break statement in DataHandler to exclude unsupported tables from processing.
  28. TYPO3com voted on patch set 7 V -1 02/22/20, 16:34 Completed build in 29s on Sat, 22 Feb, 04:34 PM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-486

  29. Claus Due wrote 02/22/20, 16:34

    Description set to "Reduction in feature footprint; removed support for DB-based definitions of content grids."
  30. Claus Due uploaded patch set 8 02/22/20, 16:35

  31. Claus Due wrote 02/22/20, 16:35

    Description set to "Fixed CGL"
  32. TYPO3com voted on patch set 8 V +1 02/22/20, 16:50 Completed build in 14m 50s on Sat, 22 Feb, 04:50 PM Test Summary: 71572 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-487

  33. Kevin Appelt commented on patch set 8 02/22/20, 19:27

    DefaultConfigurationDescription.yaml:213

    elemetns => elements
  34. Kevin Appelt commented on patch set 8 02/22/20, 20:39

    Feature-90444-NestedContentGridsFeature.rst:148

    typoscriptObjectPath (small "s" in typoscript)
  35. Claus Due uploaded patch set 9 02/22/20, 20:52

  36. Claus Due commented on patch set 9 02/22/20, 20:54

    Changes: * Allow SubContentProcessor to receive and pass-through every option that CONTENT cObject supports, and document this behavior in RST. * Dispatch FlashMessage if colPos in layout is outside of allowed range 0-99, and document this constraint in RST. * Add additional information about feature in RST, such as more PHP examples. Fixes: * Fix typo in description next to feature toggle, "elemetns" -> "elements"
  37. TYPO3com voted on patch set 9 V +1 02/22/20, 21:07 Completed build in 14m 41s on Sat, 22 Feb, 09:07 PM Test Summary: 71632 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-514

  38. Kevin Appelt uploaded patch set 10 02/22/20, 21:30

  39. Kevin Appelt commented on patch set 10 02/22/20, 21:45

    DefaultConfigurationDescription.yaml:213 Reply

    Done

    Feature-90444-NestedContentGridsFeature.rst:148 Reply

    Done
  40. Daniel Siepmann commented on patch set 10 V +1 02/22/20, 21:58

    PageTsContentGridDataProvider.php:88

    What's the reason to not use a type hint?

    ColumnNumberUtility.php:3 Unresolved

    Missing license doc block?

    ColumnNumberUtility.php:20

    Is "Flux" related to something inside of TYPO3, or still EXT:flux?

    Feature-90444-NestedContentGridsFeature.rst:73

    I guess that option does not get explained enough. That allows this grid to be available in the configured content elements I guess? So you need to add a field to allow selection, and also need to allow the CType here. Which I find ok, but maybe that needs a bit more explanation.

    Feature-90444-NestedContentGridsFeature.rst:193

    `column.items` is always an empty array. Maybe I did something wrong? Works well in backend module "Page". Also `subContent` has rendered content elements, that's also working.

    Feature-90444-NestedContentGridsFeature.rst:197

    It has to be `typoscriptObjectPath`. Also I guess people following that approach would like to skip generation in `subContent`? So maybe we could make it optional to define both options within DataProcessing? So one has to define at least one of them, and depending on which one is defined, they get filled? Would that improve performance?
  41. TYPO3com voted on patch set 10 V +1 02/22/20, 22:19 Completed build in 37m 57s on Sat, 22 Feb, 10:19 PM Test Summary: 71634 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-519

  42. Jonas Eberle commented on patch set 10 02/23/20, 01:42

    Please provide an example how to render in plain TypoScript, too.
  43. Benjamin Kott commented on patch set 10 V -1 CR -1 02/23/20, 10:53

    Initiative Feedback still available. Please wait for Rachel and Claus to sort this out.
  44. Claus Due commented on patch set 10 02/23/20, 12:29

    PageTsContentGridDataProvider.php:88 Reply

    Overrides a method from PageTsBackendLayoutDataProvider - if it was a new method I would have added hints along with return void.

    ColumnNumberUtility.php:20 Reply

    Reworded, the utility class here is lifted from Flux since that is where the colPos calculation logic was matured.

    Feature-90444-NestedContentGridsFeature.rst:73 Reply

    Elaborated in RST with a paragraph dedicated to explaining this option.

    Feature-90444-NestedContentGridsFeature.rst:193 Reply

    Caused by undefined page ID in resolved backend layout, fixed in next push.

    Feature-90444-NestedContentGridsFeature.rst:197 Reply

    Fixed. Next push contains logic in SubContentProcessor to avoid assigning either the "grid" or "content" variable if their options are configured as empty strings or zero. This allows a developer/integrator to use one or the other, or both. Default behavior is to assign both with the default variable names. This too is explained in RST in next push.
  45. Claus Due commented on patch set 10 02/23/20, 12:33

    @Jonas Eberle > Please provide an example how to render in plain TypoScript, too. Please clarify, I'm not sure I understand. If you mean for example that you want to avoid using `FLUIDTEMPLATE` and would rather use something like a `COA` with a nested `CONTENT` object then this too is possible. It does however mean: 1. You have to specify the "select.where" option on your `CONTENT` TS object. 2. The value has to be either calculated (see ColumnNumberUtility) or hardcoded. I'm not sure I can give you an example of how to calculate that value but I suspect a `USER` with `userFunc` referring to `\TYPO3\CMS\Backend\Utility\ColumnNumberUtility::calculateColumnNumberForParentAndColumn` (which takes two arguments, parent record UID and uncalculated colPos of nested grid column) could be used? Unfortunately I'm not currently aware how to pass those arguments but perhaps someone else can. If necessary I would then add this to the RST as a section "Rendering with TS without FLUIDTEMPLATE".
  46. Claus Due uploaded patch set 11 02/23/20, 12:37

  47. Claus Due commented on patch set 11 02/23/20, 12:41

    > 1. You have to specify the "select.where" option on your `CONTENT` TS object. > 2. The value has to be either calculated (see ColumnNumberUtility) or hardcoded. The goal being to provide "colPos = $calculatedColPos" to the condition that selects content records to be rendered.
  48. Claus Due wrote 02/23/20, 12:41

    Description set to "Empty "items" in Grid in FE, improve RST, make SubContentProcessor able to process+assign either or both of the target variables"
  49. TYPO3com voted on patch set 11 V +1 02/23/20, 12:50 Completed build in 13m 03s on Sun, 23 Feb, 12:50 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-553

  50. Claus Due uploaded patch set 12 02/23/20, 15:15

  51. Claus Due wrote 02/23/20, 15:16

    Description set to "Fix issue with copying of pages causing 1) db "unknown column" error and 2) child content placed in "unused" section"
  52. TYPO3com voted on patch set 12 V +1 02/23/20, 15:41 Completed build in 22m 37s on Sun, 23 Feb, 03:41 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-568

  53. Jonas Eberle commented on patch set 12 02/23/20, 16:04

    > Patch Set 10: > > @Jonas Eberle > > > Please provide an example how to render in plain TypoScript, too. > > I'm not sure I can give you an example of how to calculate that value but I suspect a `USER` with `userFunc` referring to `\TYPO3\CMS\Backend\Utility\ColumnNumberUtility::calculateColumnNumberForParentAndColumn` (which takes two arguments, parent record UID and uncalculated colPos of nested grid column) could be used? Maybe it is just me, but I don't really get how this feature sits inside the current TYPO3 page rendering universum. IMHO there are currently 2 major ways to set up the basic page rendering: 1) TEMPLATE like here https://docs.typo3.org/c/typo3/cms-fluid-styled-content/master/en-us/Installation/InsertingContentPageTemplate/Index.html 2) CONTENT with colPos constraints page.10 = CONTENT page.10 { table = tt_content select.where = colPos = 0 } I suggest to add information to the .rst what would be the recommended way what would have to be changed for these ways to use grids.
  54. Claus Due commented on patch set 12 02/23/20, 16:50

    > 1) TEMPLATE like here https://docs.typo3.org/c/typo3/cms-fluid-styled-content/master/en-us/Installation/InsertingContentPageTemplate/Index.html > > 2) CONTENT with colPos constraints > page.10 = CONTENT > page.10 { > table = tt_content > select.where = colPos = 0 > } > > I suggest to add information to the .rst what would be the recommended way what would have to be changed for these ways to use grids. Page templates and the TS that renders them are unaffected and work precisely like before. The only change is at content-level. CONTENT works the same but rendering of a content type that has nested content has to be extended in order to allow and render the sub-content and assign that as template variable (almost precisely what fluid_styled_content does, integration done by adding a "dataProcessing" entry utilising the SubContentProcessor). The use of the data processor and how you can output/render sub-content in a Fluid template is already in the RST. The thing that might be missing is related to much older practices of not using FLUIDTEMPLATE but having, for example, COA do the rendering (honestly I'm not even sure we recommend that practice anymore). There are also examples in RST how you would go about handling sub-content in for example a controller for a Extbase or other plugins. I will add a section in RST about "page rendering" and inform that no changes whatsoever are required - the only changes you'd ever need to do, would be for each specific content type that you as developer decided should support nested content. Hopefully this answers your question? Let me know if I can explain any parts of the info in RST better!
  55. Kevin Appelt commented on patch set 12 02/23/20, 16:58

    > Patch Set 12: > > > Patch Set 10: > > > > @Jonas Eberle > > > > > Please provide an example how to render in plain TypoScript, too. > > > > I'm not sure I can give you an example of how to calculate that value but I suspect a `USER` with `userFunc` referring to `\TYPO3\CMS\Backend\Utility\ColumnNumberUtility::calculateColumnNumberForParentAndColumn` (which takes two arguments, parent record UID and uncalculated colPos of nested grid column) could be used? > > Maybe it is just me, but I don't really get how this feature sits inside the current TYPO3 page rendering universum. > > IMHO there are currently 2 major ways to set up the basic page rendering: > > 1) TEMPLATE like here https://docs.typo3.org/c/typo3/cms-fluid-styled-content/master/en-us/Installation/InsertingContentPageTemplate/Index.html > > 2) CONTENT with colPos constraints > page.10 = CONTENT > page.10 { > table = tt_content > select.where = colPos = 0 > } > > I suggest to add information to the .rst what would be the recommended way what would have to be changed for these ways to use grids. This way is still valid as you can see here: https://github.com/kevin-appelt/bootstrapgrid/blob/master/Configuration/TypoScript/Page/setup.typoscript#L93 The TypoScript in the .rst is for the content element itself. Example here: https://github.com/kevin-appelt/bootstrapgrid/blob/master/Configuration/TypoScript/setup.typoscript
  56. Rachel Foucard commented on patch set 12 02/23/20, 17:07

    Hello, developers! Claus and I had a chat and I confirm that I agree with this feature, and that we’ll be working further on it's UX as a united group ;)
  57. Claus Due uploaded patch set 13 02/23/20, 17:08

  58. Claus Due wrote 02/23/20, 17:09

    Description set to "Avoid potential e_notice level message in XHR after drag-n-drop of content"
  59. Claus Due updated the commit message in patch set 14 02/23/20, 17:19

  60. Claus Due wrote 02/23/20, 17:20

    Description set to "Adjusted commit message"
  61. TYPO3com voted on patch set 13 V +1 02/23/20, 17:22 Completed build in 13m 53s on Sun, 23 Feb, 05:22 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-573

  62. TYPO3com voted on patch set 14 V +1 02/23/20, 17:37 Completed build in 16m 21s on Sun, 23 Feb, 05:37 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-575

  63. Kevin Appelt uploaded patch set 15 02/23/20, 19:10

  64. TYPO3com voted on patch set 15 V +1 02/23/20, 19:24 Completed build in 13m 18s on Sun, 23 Feb, 07:24 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-580

  65. Kevin Appelt uploaded patch set 16 02/23/20, 19:26

  66. TYPO3com voted on patch set 16 V +1 02/23/20, 19:39 Completed build in 13m 05s on Sun, 23 Feb, 07:39 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-581

  67. Kevin Appelt uploaded patch set 17 02/23/20, 19:51

  68. Claus Due uploaded patch set 18 02/23/20, 20:03

  69. TYPO3com voted on patch set 17 V -1 02/23/20, 20:07 Completed build in 15m 42s on Sun, 23 Feb, 08:07 PM Test Summary: 71333 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-582

  70. TYPO3com voted on patch set 18 V +1 02/23/20, 20:38 Completed build in 27m 56s on Sun, 23 Feb, 08:38 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-583

  71. Claus Due uploaded patch set 19 02/23/20, 21:00

  72. Claus Due wrote 02/23/20, 21:01

    Description set to "Fix translation of child when parent is already translated, fix translate-by-copy mode for child content."
  73. TYPO3com voted on patch set 19 V +1 02/23/20, 21:25 Completed build in 25m 10s on Sun, 23 Feb, 09:25 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-585

  74. Claus Due uploaded patch set 20 02/23/20, 21:32

  75. TYPO3com voted on patch set 20 V +1 02/23/20, 21:47 Completed build in 15m 32s on Sun, 23 Feb, 09:47 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-588

  76. Claus Due uploaded patch set 21 02/23/20, 21:57

  77. Claus Due wrote 02/23/20, 21:57

    Description set to "Fix copy after element in different language"
  78. Claus Due wrote 02/23/20, 21:57

    Description "Fix copy after element in different language" removed
  79. Claus Due wrote 02/23/20, 21:57

    Description set to "Fix paste after element in different language"
  80. TYPO3com voted on patch set 21 V +1 02/23/20, 22:11 Completed build in 14m 22s on Sun, 23 Feb, 10:11 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-592

  81. Georg Ringer commented on patch set 21 CR -1 02/24/20, 06:36

    LocalizationController.php:306

    what is the reason for that? either debugging then please remove, otherwise please comment why. looks strange

    PageTsContentGridDataProvider.php:2

    declare strict

    PageTsContentGridDataProvider.php:25

    should include something like grid because we have already a tsconfig provider for backend layouts. and what is actually the difference to PageTsBackendLayoutDataProvider??

    ColumnNumberUtility.php:48

    private

    BackendLayout.php:405

    some description would be nice, also should there be a guard clause with the feature flag?

    ContentGridDataProviderInterface.php:2

    declare strict

    tt_content.php:345

    not needed, defined via selectSingle
  82. Claus Due uploaded patch set 22 02/24/20, 07:05

  83. Claus Due wrote 02/24/20, 07:07

    Description set to "Adjusted to review comments from Georg"
  84. Claus Due uploaded patch set 23 02/24/20, 07:08

  85. Claus Due wrote 02/24/20, 07:08

    Description set to "Adjusted tt_content.colPos to bigint(20) to allow high calculated colPos values"
  86. Claus Due commented on patch set 23 02/24/20, 07:08

    LocalizationController.php:306 Reply

    Leftover debug assist, xdebug is unable to break on the return statement. Removed.

    PageTsContentGridDataProvider.php:2 Reply

    Done

    PageTsContentGridDataProvider.php:25 Reply

    Reworded to "adds content grids". Difference in this class: * Reads from different pageTS location. * Implements additional interface declaring getSubBackendLayoutForContentRecord Note, methods which are not declared with strict return type are inherited from parent.

    ColumnNumberUtility.php:48 Reply

    Const is for public use, declared public now.

    BackendLayout.php:405 Reply

    Done

    ContentGridDataProviderInterface.php:2 Reply

    Done

    tt_content.php:345 Reply

    Done
  87. TYPO3com voted on patch set 22 V +1 02/24/20, 07:32 Completed build in 25m 17s on Mon, 24 Feb, 07:32 AM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-594

  88. TYPO3com voted on patch set 23 V +1 02/24/20, 07:40 Completed build in 18m 17s on Mon, 24 Feb, 07:40 AM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-595

  89. Claus Due uploaded patch set 24 02/24/20, 08:33

  90. Claus Due wrote 02/24/20, 08:33

    Description set to "Removed testing fixture that was not supposed to be committed"
  91. TYPO3com voted on patch set 24 V +1 02/24/20, 08:59 Completed build in 17m 28s on Mon, 24 Feb, 08:59 AM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-600

  92. Mathias Brodala commented on patch set 24 02/24/20, 09:04

    Feature-90444-NestedContentGridsFeature.rst:53

    Is this still up to date? I cannot see most of these classes ...

    Feature-90444-NestedContentGridsFeature.rst:73

    So this configures the allowed CTypes for all sections (I'd prefer this term over "columns in rows")? That's not really useful IMO. Here EXT:content_defender does a better job so maybe we should skip this for now and instead give that extension all it needs.

    Feature-90444-NestedContentGridsFeature.rst:102

    How to allow everything then? :-)

    Feature-90444-NestedContentGridsFeature.rst:126

    Maybe stupid question: what does this whole section mean? It sounds like I will be able to select a content grid type in my content element, but then? I won't be able to maintain/fill the content grid then, right?

    Feature-90444-NestedContentGridsFeature.rst:140

    What is this for? Shouldn't everything be part of the "grid"?

    Feature-90444-NestedContentGridsFeature.rst:208

    OK, now I get what this is for. Still I think this could be merged into the grid which then needs a way for Fluid to access content by colPos/name.
  93. Claus Due commented on patch set 24 02/24/20, 09:18

    Feature-90444-NestedContentGridsFeature.rst:53 Reply

    Those classes are part of a previous, already merged feature - the RST here merely repeats a description so it will be logically grouped with this feature's description. Reason is that examples below use the hierarchy for one of the two rendering suggestions.

    Feature-90444-NestedContentGridsFeature.rst:73 Reply

    It does not, please read below: it declares which CTypes are allowed to select the content grid for use as "sub layout". It does not restrict allowed CTypes inside the grid. That is still the domain of extensions like content_defender.

    Feature-90444-NestedContentGridsFeature.rst:102 Reply

    Disallowing "*" is quite intentional. The option serves as a restriction to ensure that whomever wrote the content grid definition explicitly decided which CTypes should be able to select it (and developer would also have to explicitly choose to show the field that would allow selecting a content grid by modifying TCA of that particular type). Explanation/example is at line 123.

    Feature-90444-NestedContentGridsFeature.rst:126 Reply

    Steps to use a content grid with a CType: 1. Declare CType in "allowed_content_types" of content grid definition 2. Modify TCA to allow the "content_grid" field to be shown for that CType 3. (Editor) select the content grid using the "content_grid" field (which only shows identifiers of grids that list the CType of record being edited, in "allowed_content_types" Doing the above causes the content grid to be rendered inside the ce-body of the parent content record.

    Feature-90444-NestedContentGridsFeature.rst:140 Reply

    See below; there are two ways to render content. One is by "contentAs" which makes the template variables assigned to the template contain *the fully rendered content*. The other is with "gridAs" which allows you to iterate over rows, columns and records, and render each record individually. The latter allows further control over rendering directly from Fluid template and lets you perform checks on, for example, attributes of the column (special rendering for a given column name, etc.)

    Feature-90444-NestedContentGridsFeature.rst:208 Reply

    I would disagree; the Grid and sub-objects are solely responsible for structure and delivery of variables that are used in BE for expressing the grid in page module. I would certainly not mix frontend rendering into that. See above for reasoning behind allowing either/both of "subContent" and "grid" variables.
  94. Kay Strobach commented on patch set 24 02/24/20, 09:29

    Feature-90444-NestedContentGridsFeature.rst:73

    as the naming cause a lot of confusion i would argue for renaming "allowed_content_types" to "register_for_content_types" or "available_in_content_types" or just "for_content_types"

    tt_content.php:345 Reply

    Ack
  95. Claus Due commented on patch set 24 02/24/20, 09:31

    Feature-90444-NestedContentGridsFeature.rst:73 Reply

    Fair point. I would then rename to "for_content_types" since "content_types" is a bit generic and "register_for_content_types" is a bit long-winded ;)
  96. Claus Due uploaded patch set 25 02/24/20, 09:44

  97. Claus Due wrote 02/24/20, 09:45

    Description set to "Removed "backend layotus" from commit message subject, removed previous PS related fields from backend_layout TCA showitem, removed redundant overridden method, renamed "allowed_content_types" to "for_content_types""
  98. Claus Due commented on patch set 25 02/24/20, 09:50

    Feature-90444-NestedContentGridsFeature.rst:73 Reply

    Done
  99. TYPO3com voted on patch set 25 V +1 02/24/20, 09:58 Completed build in 13m 41s on Mon, 24 Feb, 09:58 AM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-606

  100. Jonas Eberle commented on patch set 25 02/24/20, 10:08

    > Page templates and the TS that renders them are unaffected and work precisely like before. The only change is at content-level. > > CONTENT works the same but rendering of a content type that has nested content has to be extended in order to allow and render the sub-content... Thank you, now I am on the right page. I think the .rst explains it well but I got tricked into the wrong direction by the mention backend layouts and page layout.

    Feature-90444-NestedContentGridsFeature.rst:18

    I would call this module just "page module". I think this wording is consistent with other places, e.g. https://docs.typo3.org/m/typo3/reference-coreapi/master/en-us/ApiOverview/BackendLayout/Index.html
  101. Claus Due uploaded patch set 26 02/24/20, 10:48

  102. TYPO3com voted on patch set 26 V -1 02/24/20, 10:49 Completed build in 24s on Mon, 24 Feb, 10:49 AM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-610

  103. Claus Due wrote 02/24/20, 10:50

    Description set to "Adjusted wording in RST, fixed edge case of wrong colPos in copied child content if translating in "copy" mode from language that's in "connected" mode with another language."
  104. TYPO3com commented on patch set 26 02/24/20, 10:53 Completed build in 23s on Mon, 24 Feb, 10:53 AM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-611

  105. Claus Due uploaded patch set 27 02/24/20, 10:54

  106. Claus Due commented on patch set 27 02/24/20, 11:05

    Feature-90444-NestedContentGridsFeature.rst:18 Reply

    Done
  107. TYPO3com voted on patch set 27 V +1 02/24/20, 11:09 Completed build in 14m 31s on Mon, 24 Feb, 11:09 AM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-612

  108. Claus Due uploaded patch set 28 02/24/20, 12:26

  109. Claus Due commented on patch set 28 02/24/20, 12:27

    Fixed missing button label for "+ Content" button in top of nested column, requires https://review.typo3.org/c/Packages/TYPO3.CMS/+/63390 merged to actually show the title.
  110. TYPO3com voted on patch set 28 V +1 02/24/20, 12:40 Completed build in 13m 45s on Mon, 24 Feb, 12:40 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-614

  111. Tymoteusz Motylewski commented on patch set 28 02/24/20, 12:44

    Feature-90444-NestedContentGridsFeature.rst:73

    please document this property
  112. Claus Due uploaded patch set 29 02/24/20, 12:54

  113. Claus Due commented on patch set 29 02/24/20, 12:54

    Feature-90444-NestedContentGridsFeature.rst:73 Reply

    Done
  114. Claus Due wrote 02/24/20, 12:55

    Description set to "Further documented purpose of "for_content_types" option in pageTSconfig grid definition"
  115. Jigal van Hemert commented on patch set 29 02/24/20, 13:03

    It's (again (historically there were often big, late features which had quite some issues on release day)) way too short before feature freeze to include a feature with such impact as this. There is no UX/UI decided yet, so it's by far not stable enough to ship (even if it's marked as an experimental feature).

    BackendLayoutView.php:323

    ColumnNumberUtility::MULTIPLIER instead of 100?

    DataHandler.php:3172

    $nestedContentGridsEnabled already contains the value of this feature switch, doesn't it? So, it could actually be an if-else statement instead of these two ifs?

    Feature-90444-NestedContentGridsFeature.rst:104

    This range can easily be too limited for large, multi-site installation with loads of custom content elements. Especially if colPos values have subranges to separate various sites or for other reasons.

    Feature-90444-NestedContentGridsFeature.rst:277

    There is no limitation on the uid of a record other than the type of the database field and the PHP variable. This feature suddenly reduces this range without any checks in the rest of the core. If the allowed range of 0-99 is expanded the uid of the parent record is even more limited. Some installation have content stagings and might use ranges in UIDs to separate content from different sources. If in the furture the UID is replaced by a different system (UUID for example) this connection between colPos and uid will become another limitation and a local UID number needs to be kept to allow such calculations as proposed here to keep working.

    ext_tables.sql:119

    I'm not sure what the impact is with doctrine queries. According to the doctrine documentation https://www.doctrine-project.org/projects/doctrine-dbal/en/2.10/reference/types.html#bigint it will use PHP string types to represent bigint values because 32-bits PHP systems will not be able to store them in an integer variable. PHPs internal type juggling might use floats on those systems which would lead to unexpected behaviour. The only place in the core where 'bigint' is currently used is the file size in FAL,which is kind of non-critical. I guess the minimum requirement of TYPO3 should be a 64-bit PHP version.
  116. Claus Due commented on patch set 29 02/24/20, 13:19

    BackendLayoutView.php:323 Reply

    Indeed! Good catch, will change this.

    DataHandler.php:3172 Reply

    Definitely, I neglected to refactor this after extracting previous logic to a dedicated method. I will condense this to a single condition.

    Feature-90444-NestedContentGridsFeature.rst:104 Reply

    The limitation is chosen based on both practical experience from Flux (which is where the calculated colPos idea comes from) and no users reported having any projects that needed more than 100 possible colPos values. It could also be argued that colPos values only need to be unique for the individual backend layout, and will not conflict with colPos values of content grids. E.g. each backend layout can have 100 columns and each content grid can have 100 columns without causing any colissions with either each other or other backend layouts / content grids. Lastly, sites which do use that many different columns in backend layouts are unlikely to enable the feature flag and therefore would not be limited to 100 columns in backend layouts.

    Feature-90444-NestedContentGridsFeature.rst:277 Reply Unresolved

    I'm not sure I understand this fully. There are no restrictions on uid imposed by this feature. Only colPos is restricted to a maximum of 100 columns in each grid definition. If the core at some point switches to UUID it will be necessary to come up with many, many other solutions as well - in the larger perspective, this relation should be the smaller of problems. There are AFAIK no plans, close or far in the future, to switch to UUID.

    ext_tables.sql:119 Reply

    Field sys_file.size also is declared as bigint and would be equally affected. I certainly don't consider FAL's filesizes non-critical as FAL is at this point a requirement of multiple core functions and is recommended practice for third party integration. If 64 bit isn't already a requirement then I would agree it should be - but that is out of scope for this patch.
  117. Claus Due uploaded patch set 30 02/24/20, 13:19

  118. Claus Due wrote 02/24/20, 13:20

    Description set to "Adjusted to review comments from Jigal"
  119. TYPO3com voted on patch set 29 V +1 02/24/20, 13:28 Completed build in 28m 58s on Mon, 24 Feb, 01:28 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-620

  120. Jonas Eberle commented on patch set 30 02/24/20, 13:29

    ext_tables.sql:119 Reply

    Although it might sound anachronistic, don't take 64bit PHP for granted please. DomainFactory, among others, have their php-fpm-stack running as a 32bit executable! (If you want to quickly check others: php -r 'echo PHP_INT_SIZE === 4? "32bit": "64bit";') Can we avoid bigint here?
  121. Claus Due commented on patch set 30 02/24/20, 13:35

    ext_tables.sql:119 Reply

    Avoiding bigint would cause potential breakage for sites with sufficiently high UID values in tt_content. While it is technically possible to avoid it is not advisable, imho, as it introduces a hidden boundary. The same applies to the sys_file.size field - so there is current practice of using bigint. See also Jigal's note about doctrine conversion of 64bit types for use on 32bit platforms.
  122. TYPO3com voted on patch set 30 V +1 02/24/20, 14:05 Completed build in 40m 31s on Mon, 24 Feb, 02:04 PM Test Summary: 71640 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-625

  123. Kay Strobach commented on patch set 30 CR +1 02/24/20, 15:28

    code looks good, previous iteration would also get a +1 for verified
  124. Tymoteusz Motylewski commented on patch set 30 02/24/20, 15:41

    PageTsContentGridDataProvider.php:70

    doc type missing?

    PageTsContentGridDataProvider.php:90

    lets have a type for the param

    PageTsContentGridDataProvider.php:111

    this would be array|null instead of mixed, right?

    PageTsContentGridDataProvider.php:113

    lets have a type for the params and maybe for the return type

    ColumnNumberUtility.php:60

    type hints plz

    ColumnNumberUtility.php:65 Unresolved

    type hints plz

    ColumnNumberUtility.php:73 Unresolved

    type hints plz

    BackendLayoutView.php:105

    where is this function called from? Shouldnt it return sth, or are $parameters passed as reference?

    BackendLayoutView.php:122

    : void

    DataHandler.php:2993

    :void

    DataHandler.php:3202

    :void what types $commandvalue can take? at later point there is a check "if ($commandValue > 0) {" so $commandValue has to be numeric?

    DataHandler.php:4560

    would it make sense to extract it to a method?
  125. Jo Hasenau commented on patch set 30 CR +1 02/24/20, 15:42

    While I still dislike they way how values of colPos are based on calculation, I have to admit, that it's working for almost any realistic scenario.
  126. Claus Due commented on patch set 30 02/24/20, 15:48

    PageTsContentGridDataProvider.php:70 Reply

    Is this required for methods that are strict-typed throughout? If so I will of course add them as there are other occurrences in the patch.

    PageTsContentGridDataProvider.php:90 Reply Unresolved

    The method is inherited, I'm not sure it would be wise to make it strict.

    PageTsContentGridDataProvider.php:111 Reply Unresolved

    Correct, but method is inherited and doc block is a duplicate of parent.

    PageTsContentGridDataProvider.php:113 Reply Unresolved

    This too is an inherited method, see my question above.

    ColumnNumberUtility.php:60 Reply Unresolved

    Coming right up (x3)

    BackendLayoutView.php:105 Reply

    Method is no longer used, will remove.

    BackendLayoutView.php:122 Reply

    Ack

    DataHandler.php:2993 Reply

    Ack

    DataHandler.php:3202 Reply

    Command value can be array or integer; is integer for all cases that in this method triggers behavior, but cannot be strict typed without moving both feature flag checks, table checks, command checks and loops to multiple other places in code. For isolation I chose to allow mixed type here.

    DataHandler.php:4560 Reply Unresolved

    That can be done, but unless you insist on doing it now, I would prefer to do that as a followup.
  127. Tymoteusz Motylewski commented on patch set 30 02/24/20, 16:01

    PageTsContentGridDataProvider.php:70 Reply Unresolved

    you're right, not required.

    DataHandler.php:3202 Reply

    Ack
  128. Daniel Siepmann commented on patch set 30 02/24/20, 16:57

    I can end up in mixed mode when translating in backend: 1. Translate via "Translate" (connected mode) 2. Delete all content elements in translation 3. Translate via "Copy" (free mode) Still everything looks ok visually. One does not see mixed content. Content elements in nested grid are localized to backend colPos instead of localized grid: 1. Create grid (same colPos as backend layout) 2. Add content to grid 3. Copy grid into itself 4. Translate using "copy" free mode

    Feature-90444-NestedContentGridsFeature.rst:83 Unresolved

    For backend layouts a message is rendered and a different background is used when no colPos is defined. For content grids a colPos is calculated and displayed, that's expected? Not providing colPos at all, results in auto generated title, providing an empty colPos results in defined title with auto calculated colPos.

    Feature-90444-NestedContentGridsFeature.rst:110 Unresolved

    Can we clearify where these colPos values should be avoided? Should they be avoided only in ContentGrid or also backend layouts?

    Feature-90444-NestedContentGridsFeature.rst:221 Unresolved

    The column (GridColumn::class) has does not have the configured colPos, but calculated colPos. That way it is not possible to iterate and use the generated content in `subContent.{column.columnNumber}`. Would it be possible to add the configured colPos in addition to calculated one? That would allow to combine both options of DataProcessor to take care of rendering, and still provide structure. But no must have for me.

    Feature-90444-NestedContentGridsFeature.rst:295 Unresolved

    This is an php example, no typoscript.
  129. Claus Due commented on patch set 30 CR -1 02/24/20, 17:23

    With a slightly heavy heart, I'm sorry that I have to inform you all, that: * This feature will not make it into 10.3 regardless of possibly obtaining the necessary verification votes this evening. * I've spoken to the core team members about a possible exemption from feature freeze, meaning this feature would be allowed to merge to master, assuming that it is able to gain the necessary acceptance through further testing and review. To all the people who helped code, test and review this - I can't thank you enough for making the effort to shoot for inclusion in 10.3, it's been some very long working days, but we are simply out of time and this is one feature I do not want to "force" into a released version without broader acceptance. The good news is that this gives myself and the structured content initiative more time to communicate about the goals and intentions of the feature, aiming to produce an official typo3.org article for public consumption. I'm terribly sorry we didn't make it in time - but equally happy that there is a (conditional) exemption from feature freeze that *still* allows this to happen as part of the v10 LTS release. We choose the safer, more cautious road toward LTS! (placing -1 for the time being, until 10.3 has been released - just in case)
  130. Jigal van Hemert commented on patch set 30 02/24/20, 20:04

    Feature-90444-NestedContentGridsFeature.rst:104 Reply

    People use all kinds of rules for the meaning of a number. An installation with 5 subsites may use ranges for colPos numbers. E.g. 0-20 for site A, 21-40 for site B, and so on. Or colPos numbers are the same as some other ID number. There is no technical limitation for colPos numbers, only a rule in the code of this patch that dictates the limitation.

    ext_tables.sql:119 Reply

    If sys_file.size is larger than a 32-bit int and used in calculations it will be type juggled into a float. As far as I could find the file size is only used to display the size. Huge numbers are usually displayed with a limited precision and a float representation of a huge number will have enough precision to not notice any difference. This is non-critical to me. The UID of any record needs to be unique and must not change it's value between database and PHP. A UID is critical in my opinion.
  131. Claus Due commented on patch set 30 02/24/20, 20:32

    Feature-90444-NestedContentGridsFeature.rst:104 Reply Unresolved

    > There is no technical limitation for colPos numbers, only a rule in the code of this patch that dictates the limitation. Forgive me Jigal, but I still don't understand the objection. Yes, there is a limitation for range of colPos. But this limitation is not imposed on UIDs of any kind, nor is there any colission (numbers are unique absolutely only inside the columns contained within one definition). The worst-case scenario for a site like this (which I assume would have to be an existing site with a massive setup) is that it can't take advantage of the feature and would need to leave the feature flag OFF. By the way, this patch precisely aims to provide an alternative to very large and many-columned definitions. Personally I would evaluate these two use cases as mutually exclusive, but perhaps I'm overlooking something?

    ext_tables.sql:119 Reply

    This would mean the maximum UID of tt_content records before encountering issues with calculated colPos on a 32bit system would be 21,474,836 (correct me if I'm wrong, please. 2147483647/100). This limit would be the same regardless of int or bigint (again, correct me if I'm wrong). And it would mean that the only case that can trigger any issue on 32 bit systems with bigint, would be a tt_content record having a UID of >=21474837 into which child content is inserted. The effect would be either that this record's child content would not be resolved (appears empty) or a type error from PHP (depending on how we treat the case). I'm not sure what the effect would be with int(11) and a tt_content record exceeding the colPos>2147483647 boundary. Perhaps you know this and can inform me. I could implement an exception for this rather edge case, checking if calculated colPos would exceed PHP_INT_MAX, possibly combined with a message that a 64bit system would not be affected, and/or instructing to disable the feature. What would be your thoughts on this? Is it even remotely conceivable any site would reach the required UID number in tt_content? Is an exception preferable?
  132. Jonas Eberle commented on patch set 30 02/24/20, 21:49

    ext_tables.sql:119 Reply

    I would like it best if we could not deal with colPos at all but rather use a concept like 'parent'. Maybe explicitly set colPos=-1 for subcontent. Or keep the same colPos as the parent to allow for efficient fetching of content (and then a PHP traversal to reorder it).
  133. Claus Due commented on patch set 30 02/24/20, 22:05

    ext_tables.sql:119 Reply

    Not using colPos has severe implications for everything from UI over JS to DataHandler and grid definitions; colPos was chosen for this iteration because the feature will then be able to leverage large amounts of existing code without need for change. Same colPos as parent would demand that every instance that reads page content (TypoScript CONTENT objects etc.) would have to add a secondary condition. The chosen strategy has been proven in production for about two years, as it is the parent/child relation strategy used in Flux. It allowed the number of hooks and overrides required to be reduced to around 10% of what they initially were, when Flux used parent+column fields in DB, which should give an indication of how much difference there is in implementation scope between colPos or parent+column relation. It might not be ideal but it is a good starting point in terms of reusing existing logic of many other contexts. If you want to help us design the further iterations I invite you to join the structured content initiative! There will also be public decision discussions on decisions.typo3.org (though perhaps not for this first iteration / experimental first stage).
  134. Jonas Eberle commented on patch set 30 02/25/20, 08:15

    ext_tables.sql:119 Reply

    What about colPos=-1? I think it would be good to avoid squeezing 3 distinct data into 1 column with modulo and offset. This is the memory layout of a 2dimensional array. It is cool that this is mathematically possible but complex, too. I would try to avoid it. Adding multiple columns instead can't be too problematic. Thank you for inviting me. I hope to take my concern although I am not part of the initiative.
  135. Claus Due commented on patch set 30 02/25/20, 08:38

    ext_tables.sql:119 Reply Unresolved

    Having the same colPos value for all nested content is the same challenge - it has so many implications for other parts of the system that it wouldn't work without extensive overrides and replacements (essentially, it would require a secondary mode of operation for everything dealing with positioning of content elements). I promise that all alternatives have been investigated fully, for the past 5+ years, in various third party extensions. The unique colPos and calculation approach is by far the one that has the fewest implications. It may look deceptively simple but when you get into it you find thing after thing that wouldn't work *unless* the colPos values are unique.
  136. Jo Hasenau commented on patch set 30 02/25/20, 09:19

    > Patch Set 30: > > (1 comment)
  137. Jo Hasenau commented on patch set 30 CR -1 02/25/20, 09:22

    The colPos calculation is actually my major concern with this patchset too. We have got several approaches to a solution, but squeezing 3 different meanings into the calculated value of colPos feels definitely wrong to me, while I have to admit that it looks easy to do. The -1 used by Gridelements feels less wrong and would actually be working properly, if the core would not reset colPos to unsigned while people are upgrading having Gridelements uninstalled. Still a clean solution should not provide a colPos value for child elements at all, since their position is related to the parent element only, so they don't need to "know" about anything else. There should be 2 different fields containing those coordinates in the multidimensional array of the "content tree" just as there are 2 different fields to position content on pages. For pages we've got pid and colPos, so we should not abuse that for nested content. colPos should be set to NULL for child elements to remove them from the first level of content queries. And there should be a distinct parent and a column field for those child elements to determine their position. The structured content initiative is actually working on a concept that would completely remove positions and coordinates from data records, so we should already provide a solution that has been normalized as far as possible if we go for an in between solution now.
  138. Claus Due commented on patch set 30 02/25/20, 09:49

    > The -1 used by Gridelements feels less wrong and would actually be working properly, if the core would not reset colPos to unsigned while people are upgrading having Gridelements uninstalled. That is most definitely not my experience (the -1 working thing), as that has multiple side effects in the output and JS as far as I've tested, but we can have a call about this!
  139. Jo Hasenau commented on patch set 30 02/25/20, 10:40

    > Patch Set 30: > > > The -1 used by Gridelements feels less wrong and would actually be working properly, if the core would not reset colPos to unsigned while people are upgrading having Gridelements uninstalled. > > That is most definitely not my experience (the -1 working thing), as that has multiple side effects in the output and JS as far as I've tested, but we can have a call about this! Just to clarify this: I know that with current core only methods there are side effects with -1 and maybe even with NULL as a value for colPos. The point is, that we should not go for a hacky workaround to those side effects, while i.e. Gridelements is working fine after providing a solution for those normalized DB values.
  140. Rachel Foucard commented on patch set 30 02/25/20, 13:08

    > The structured content initiative is actually working on a concept that would completely remove positions and coordinates from data records, so we should already provide a solution that has been normalized as far as possible if we go for an in between solution now. Hi guys, you (Jonas, Claus, Joe, etc.) seem to be very inspired right now to debate and argue the possibilities. We have started a google document, it would be nice to integrate your ideas and comment on the ones that are already there. If you want me to share this link with you, just tell me on slack.
  1. Claus Due commented on patch set 30 03/01/20, 22:56

    Affected by: * https://forge.typo3.org/issues/90610 which causes duplicates of an already localised content element in the localised language, when copy-pasting a content record to a third language. * https://forge.typo3.org/issues/85013 which causes TYPO3 to not move a localised version of a record if the original language record is copied and pasted to the top of a column (which includes the top of a nested column). Neither of these symptoms are caused by, or fixed by, this patch. Fixing the original cause for the two problems above will also fix them for this patch.
  2. Claus Due commented on patch set 30 03/01/20, 23:15

    Observed bug (to be fixed): * When translating with "free" mode and choosing a source language that is itself a translation performed in "connected" mode, any children of a copied parent element have an incorrect colPos value in database (uses parent UID of original language version, should use UID of freshly copied parent).
  3. Claus Due uploaded patch set 31 03/01/20, 23:18

  4. TYPO3com voted on patch set 31 V -1 03/01/20, 23:18 Completed build in 25s on Sun, 1 Mar, 11:18 PM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-904

  5. Claus Due wrote 03/01/20, 23:18

    Description set to "Moved DataHandler changes to a hook subscriber for cleaner separation when feature toggle is OFF"
  6. Claus Due uploaded patch set 32 03/01/20, 23:24

  7. TYPO3com voted on patch set 32 V -1 03/01/20, 23:25 Completed build in 23s on Sun, 1 Mar, 11:25 PM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-905

  8. Claus Due wrote 03/01/20, 23:25

    Description set to "Fix CGL and completely restore DataHandler including bad phpdoc"
  9. Claus Due uploaded patch set 33 03/01/20, 23:28

  10. TYPO3com voted on patch set 33 V +1 03/01/20, 23:39 Completed build in 10m 53s on Sun, 1 Mar, 11:39 PM Test Summary: 72110 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-906

  11. Jan Delius voted on patch set 33 V +1 03/18/20, 08:27

  12. Andreas Kienast commented on patch set 33 03/19/20, 06:09

    BackendLayout.php:456

    I don't think so
  13. Claus Due uploaded patch set 34 03/19/20, 09:51

  14. Claus Due commented on patch set 34 03/19/20, 09:51

    BackendLayout.php:456 Reply

    Done
  15. TYPO3com voted on patch set 34 V +1 03/19/20, 10:15 Completed build in 14m 10s on Thu, 19 Mar, 10:15 AM Test Summary: 72144 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-1438

  16. Claus Due uploaded patch set 35 03/24/20, 13:19

  17. TYPO3com voted on patch set 35 V -1 03/24/20, 13:30 Completed build in 10m 41s on Tue, 24 Mar, 01:30 PM Test Summary: 10 of 73559 failed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-1668

  18. TYPO3com commented on patch set 35 03/24/20, 14:05 Completed build in 57m 03s on Tue, 24 Mar, 02:05 PM Test Summary: 16 of 73559 failed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-1668

  19. Benni Mack uploaded patch set 36 03/26/20, 17:02

  20. TYPO3com voted on patch set 36 V -1 03/26/20, 17:02 Completed build in 22s on Thu, 26 Mar, 05:02 PM Test Summary: No tests found Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-1781

  21. Georg Ringer commented on patch set 36 CR -1 03/26/20, 17:35

    LocalizationController.php:270 Unresolved

    Doesn't looks good
  22. Benni Mack uploaded patch set 37 03/26/20, 18:36

  23. TYPO3com voted on patch set 37 V -1 03/26/20, 18:47 Completed build in 11m 00s on Thu, 26 Mar, 06:47 PM Test Summary: 73061 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-1788

  24. Claus Due commented on patch set 37 03/27/20, 12:29

    Contingency planning in case it is demanded that this feature be delivered as extension instead of core-native implementation. These are mostly notes for myself if/when the steps become necessary. Changes which are not commented on are possible to implement through over means which do not require overrides. The five items listed below would need to be addressed for this approach to be a sensible alternative (e.g. one that doesn't require XCLASS or extensive overrides which risk core-version-dependent incompatibilities).

    LocalizationController.php:265 Unresolved

    1) This method signature and the calling would need to be adapted to allow passing records and parameters to this function. Then a hook or implementation delegate would be required to change the way column consistency is extracted from the flat records and parameters.

    BackendLayout.php:135 Unresolved

    2) BackendLayout would need to be adapted with at the very least a method of assigning arbitrary values (parent layout, flag if structure is for content-level usage, etc.). A hook or implementation delegate would need to be created in the setConfiguration (and/or setStructure method, see https://review.typo3.org/c/Packages/TYPO3.CMS/+/63918.)

    BackendLayout.php:387 Unresolved

    3) It might be necessary to add an implementation delegate or hook to mutate colPos values on demand from external consumers. Technically it may be possible to avoid such a delegate or provide it without the necessity for a core change (but it would cause less duplication and more streamlined consumption if it is in core).

    BackendLayoutView.php:320 Unresolved

    4) This would be more of a challenge. Problem: core uses itemsProcFunc to generate values for colPos selector field in tt_content. But itemsProcFunc is singular and mutually exclusive: overriding it causes the original itemsProcFunc to disappear. On a general level, itemsProcFunc should be possible to define as an array of itemsProcFunc entries with each one being executed and able to modify the items. If we cannot change the itemsProcFunc the result might be either inconsistent or prone to problems in the future if/when the core's native itemsProcFunc is changed.

    Record.html:13 Unresolved

    5) This is another sensitive area: it may be a solution to call upon an optional partial template in this place, making it possible to add this partial (with these three lines of Fluid code) from an extension. The core itself would not provide any partial for this but would render it with "optional=1" to allow third parties to provide it without overriding the entire Records.html partial file.
  25. Oliver Hader abandoned the change 04/27/21, 18:39

    Seems like this change is stalled since over a year now. I'm adjusting issue state as well...