47 views
# Translation Handling Initiative<br>Team Meeting Minutes [← Back to the overview](https://notes.typo3.org/s/f3ae8fZSD) - **Date:** 2026-08-28<br> - **Where:** [Slack Huddle](https://app.slack.com/huddle/T024TUMLZ/C05D7UF1L8M) - **Participants:** - André Buchmann - Eric Harrer - Martin Clewing ## Topic 1: Translation Handling Awareness and the Current State The participants discussed how broadly translation behavior affects TYPO3 projects and how difficult it remains for developers, integrators, and customers to recognize the underlying constraints before they encounter them in practice. Martin referred to several recent project discussions about translations, fallback behavior, and international corporate requirements. Eric emphasized that multilingual content remains an important TYPO3 strength and argued that the initiative's findings should receive wider attention. Eric described the Current State document as a practical knowledge source for both newcomers and ongoing Core work. It brings together the initiative's vision, known problems, and related patches, and it had already helped him compare a new parent-child language regression with earlier work. The team considered broader communication useful, potentially including an article on typo3.org, although no publication plan was decided. ## Topic 2: Language-All Comparison View Patch Eric demonstrated [Gerrit change 95475](https://review.typo3.org/c/Packages/TYPO3.CMS/+/95475), which renders a content record with `sys_language_uid = -1` in every language column of the Layout module's Language Comparison view. The patch changes only backend representation and interaction; stored records are neither created, copied, nor migrated. Because one record now appears several times, the implementation assigns distinct DOM identifiers to its occurrences, keeps the deep-link identifier on the Default-Language occurrence, avoids treating the shared record as an inconsistent translation in Mixed Mode, and counts a hidden shared record only once. The participants verified that moving or pasting a language-specific record after a Language-All occurrence uses the language of the displayed column rather than assigning `-1`. They also identified details that still required review, including the module reload used to synchronize all visible occurrences and the JavaScript logic for the hidden-element counter. The demonstration exposed the existing limits of sorting one shared database record across independently ordered Free-Mode columns. Moving the Language-All record inside a Free-Mode column changes its single sorting value and can therefore disrupt its position in Connected-Mode columns. The team favored disabling that move operation in Free-Mode columns and requiring editors to move the surrounding language-specific records instead. The restriction and its reason should be made understandable in the feature description. Copying a Language-All record inside a Free-Mode column also still required clarification because the demonstrated copy adopted the target column's concrete language. Eric marked the change as WIP so these findings could be incorporated before general review. ## Topic 3: New Inline Child Language Inheritance The team reviewed an early implementation for the regressions reported in [Forge issue 110497](https://forge.typo3.org/issues/110497) and [Forge issue 110498](https://forge.typo3.org/issues/110498). The reports concern newly created children of unconnected non-default-language records when the child language is not submitted as a normal form field. In the accompanying initiative-channel exchange, Johannes clarified that deliberately choosing a different concrete language for a parent and child can be valid. The defect is therefore the implicit fallback to language `0` when no deliberate choice was submitted, not the general possibility of different parent and child languages. Johannes also stated that he had not assigned the issues to anyone. Eric had analyzed the reports together with previously discussed parent-child patches and generated an initial `DataHandler` implementation. If incoming data for a new inline or file child contains no explicit language, the draft derives the language from its parent. It resolves a parent from the current data map, including recursively nested new inline records, or reads an already persisted parent from the database. An explicitly submitted child language retains priority, the logic is restricted to newly created records, existing children are not converted, and no additional database write is introduced. The participants considered the initial structure plausible in a brief review, but not ready for final review. They wanted the comments shortened and aligned with Core style, the commit message clarified, the tests examined in detail, and the actual regression verified through a practical before-and-after test. They also left open whether all helper methods belong directly in `DataHandler` or should be moved to a dedicated service. Eric intended to perform that cleanup and runtime verification before uploading the draft as a clearly marked WIP change for wider review.