mirror of
https://github.com/storytold/effectcraft.git
synced 2026-10-27 03:28:22 -06:00
FeiLiuEM
2fbcc6b59a
Windows ships a UI variant of its Japanese and Chinese families (`Meiryo`/`Meiryo UI`, `Yu Gothic`/`Yu Gothic UI`, `MS Gothic`/`MS UI Gothic`, `Microsoft YaHei`/`Microsoft YaHei UI`, `Microsoft JhengHei`/`Microsoft JhengHei UI`) whose kana and kanji are drawn for the screen, and its own link table for `Segoe UI` picks the UI one for everything it draws. The interface should ask for it - but the same lists feed `fonts::fallback_for`, which shapes composed frames, so reordering them would change how existing projects render. The two orders are therefore separate: this leaves `crates/text/src/fonts.rs` exactly as it is and gives the interface its own candidate list in `install_script_fallbacks`, which falls back to the shared choice when the platform has no UI variant, none is installed, or it does not cover the probe. `PMingLiU` is in the interface list for its proportional spacing, not as a sibling of `MingLiU`: it is MingLiU's proportional relative. The rendered order is pinned by a test now: the Japanese list keeps `Yu Gothic` ahead of `Yu Gothic UI`, and no UI name appears in the lists a document renders with. `the_interface_asks_for_the_ platforms_ui_faces` pins the interface list and checks that every other script the install probes has none, so a script that draws correctly in any face keeps the shared order. Review fixes: the locale is mapped once (the call passes the language, the helper maps it), the new test sits after the #394 test it had split from its doc comment, and the fonts.rs test asserts instead of panicking inside a closure. Rebased on current main; #1395 has not landed yet, so there is nothing to rebase after it.