Fixes#4030
On a small screen (an 11-inch 1366x768 panel) the system can report a very large display scale. The Auto interface size followed it blindly, so the window and ribbon came out far bigger than the screen and File > Options, where the size is set, could not be reached.
What this changes:
- Auto interface size is now capped so a window of at least 900x460 points still fits the screen (about 152% on 1366x768). Scales that already fit (for example 1920x1080 at 100-150%, or 1366x768 at up to 150%) are unchanged. If the screen size is unknown there is no cap.
- A size chosen in File > Options is still taken as is. If a large size pushes the options out of reach, Ctrl+Shift+0 (Cmd+Shift+0 on Mac) returns to Auto from anywhere. Ctrl+Alt+0 is not used because it is AltGr+0 on some layouts and would type a character.
- `ui.interfaceSize` and the "Auto (n%)" label in Options report the capped value. The cap is never saved to `ui.json`.
- Tests: the cap for several screen sizes, the shortcut, AltGr+0 still typing, and the reported percent. Docs: ui-parity, control-protocol, ROADMAP progress log.
Not covered:
- The root cause of the oversized scale (X11 DPI detection) is handled by #721 and #3240. This change is separate and works with or without them.
- A manually chosen large size can still push controls off screen, but there is now a way back.
- Not tested on a real 1366x768 screen.
`cargo xtask ci` passed locally on the pushed code; GitHub CI has not run yet.
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
* **New Features**
* Auto interface sizing now scales down on smaller displays to keep a window of at least 900×460 points within the screen. Sizes selected in Options remain uncapped, and the minimum Auto scale is 100%.
* Press Ctrl+Shift+0 (⌘⇧0 on Mac) to restore Auto sizing. When Auto sizing is capped, the displayed zoom percentage reflects the screen-fit limit; the cap does not change the saved size.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
`cargo fmt --check` passes on CRLF (verified: exit 0, empty output), so the gate cannot see a commit that rewrites line endings. `#6537` did exactly that: 50 files, 35,102 lines, all CRLF, while declaring a one-line test change.
## Change
Adds the normalisation the repository lacks — `.gitattributes` only pins `craft-i18n.rev` today:
```
# Text lives in the repository with LF. A Windows checkout once committed 50 files with CRLF
# (#6537) and `cargo fmt --check` still passed, so the gate could not see it. Normalise on add.
* text=auto
craft-i18n.rev text eol=lf
```
`craft-i18n.rev` stays last so the specific rule still wins.
## Dependency
The repair landed first: #6541 merged as `eae8363e`, and `git grep -lI $'\r'` on that head finds **0** text files. This branch merges that `main` in `01617884`, so the rule now applies to a tree that is already LF.
`text=auto` is not an absolute enforcement gate. It normalises text on add, and it does not rewrite a blob already stored with CRLF: plain `git add` on such a file leaves `git ls-files --eol` at `i/mixed`, and only `git add --renormalize` rewrites it to `i/lf`. That is why the repair had to land first, and why this is a guard for future commits, not a fix for the past.
## Why it is safe
- Every text file in `main` is already LF (`git grep -lI $'\r'` = 0).
- No `.bat`/`.cmd` or CRLF fixture exists; the `.ps1` files under `packaging/windows/` are already LF in the repository.
- `cargo fmt --check` clean and `cargo test -p wordcraft-engine --lib` 1460 passed, unchanged.
## Why two PRs
#6541 was the repair; this is the guard, and it is a repository policy change rather than part of the fix, so it is separate and optional. If a repo-wide rule is more than you want, the narrower form is one pattern per line:
```
*.rs text eol=lf
*.toml text eol=lf
*.md text eol=lf
*.properties text eol=lf
```
Thanks for the syntax correction: the single-line fallback I wrote earlier was not valid attribute syntax.
@FredApps — the dependency is settled and this branch is on the new `main`. Ready for approval when you have a moment.
Register Latin Bosnian as an interface language, with catalog-backed locale, plural and font regressions. Preserve the existing system-locale detector; POSIX script-modifier handling is outside this change.
Original contribution by @nine-codes.
* UI: open font dropdown and select font name on click (#5914)
* UI: select font name only on opening dropdown and scope to search (#5914)
* UI: refresh highlight from current font whenever combo opens (#5914)
* UI: restore font order and match enter ordering for highlight (#5914)
* UI: address review comments on font dropdown interactions and docs (#5914)
* UI: handle system font action in dialog font combos and test focus selection (#5914)
* UI: reset highlight on open and preserve query on scrollbar focus (#5914)
* UI: drop dialog routing changes to keep scope narrow (#5914)
* UI: test genuine blur and refocus while font dropdown is open (#5914)
* chore: cut over image codecs and zip to craft-libs, fix main test regressions
* tests: store ODS mimetype and use standard ODF namespaces in mailings test
* Remove unused quick-xml workspace dependency
* fix: use standard ODF namespaces in file_dialogs test and preserve trailing line in markdown import_on_with_lines
* fix(geom): use png_chunks for PNG dimensions and p.fits(budget) in decode_for_drawing
- Wire craft-wordmodel, craft-docx, craft-rtf (doc), and craft-doc-binary into workspace
- Delete ~26,300 lines of app-owned DOCX parser/serializer in crates/docx
- Implement zero-copy media transfer and FitState custom XML round-trip via craft_wordmodel bridge
- Delete ~5,400 lines of app-owned RTF parser/serializer in crates/formats/src/rtf
- Delegate RTF import, export, parse, and esc to craft-rtf
- Add cfb_guard DIFAT validation and float sanitization on model bridge
- Net reduction: 31,697 deletions, 818 insertions (-30,879 LOC)
#6537 declared a test change but committed 50 text files with CRLF, which
`cargo fmt --check` does not catch. Restore the bytes from c37baf76: no text
file carries CRLF, and the tree is identical to that commit again.
Refs #6537, #6539.
The #5858 early return kept page anchoring for any object that does not
move with text, which includes a body frame that came from an .odt with
text:anchor-type="page". Moving it then skipped the paragraph placement
and the save wrote the page anchor and number back, contradicting #5705's
an_odt_frame_moved_from_its_page_loses_the_page_number.
Only header/footer objects (and drops on a page with no anchoring text)
keep page anchoring now, which is what #5812 needed. The body test that
pinned the opposite is updated to the #5705 rule.