render, shaping: let a caller pin the font files a frame is drawn with #466

Merged

natsukium merged 1 commit from render-pinned-fonts into main 2026-09-26 00:45:07 +09:00

Conversation 0 Commits 1 Files changed 5 +161 -7

natsukium commented 2026-09-26 00:31:18 +09:00

Owner

Copy link

Closes #441 (the pinned-fonts half; the offscreen target and readback landed in #465).

Why

Font selection always scanned the system, so a recording or screenshot rendered with different faces on two machines and a golden test could not hold its pixels. felis-reel needs to name the exact font files.

What

• RendererConfig::font_files: when non-empty, every face lookup (primary, bold/italic styles, fallback walk) runs auto_discover's discovery over these files instead of the system scan. It is read by the warm-up thread, the constructors and reload_font alike, so a reload cannot bring system faces back, and font features stay the config's as in a window.
• FontStack::discover_in_files: reads each file into memory (not mmap: the mapping's safety rests on immutable system font roots, and a caller's file can be truncated underneath it) and rejects a file with no face, which fontdb otherwise accepts silently.
• felis-client passes an empty list; its behavior is unchanged.
• docs/explanation/rendering/text-shaping.md: "Font fallback" and "Font loading" describe the pinned path.

A prebuilt FontStack handed to the constructor (the issue's proposal 3) was rejected: start_fonts, reload_font and the config's feature list could each reintroduce faces or features the stack was not built with.

Verification

• just check green.
• New tests: discovery over the Monaspace Neon test files ignores a system fallback; a missing file and a non-font file each report FontPath; a warm-up stack is discarded when font_files changes.
• pi review (sol): round 1 blocked the prebuilt-stack design; round 2 on this design passed with no findings.

Closes #441 (the pinned-fonts half; the offscreen target and readback landed in #465). ## Why Font selection always scanned the system, so a recording or screenshot rendered with different faces on two machines and a golden test could not hold its pixels. felis-reel needs to name the exact font files. ## What - `RendererConfig::font_files`: when non-empty, every face lookup (primary, bold/italic styles, fallback walk) runs `auto_discover`'s discovery over these files instead of the system scan. It is read by the warm-up thread, the constructors and `reload_font` alike, so a reload cannot bring system faces back, and font features stay the config's as in a window. - `FontStack::discover_in_files`: reads each file into memory (not mmap: the mapping's safety rests on immutable system font roots, and a caller's file can be truncated underneath it) and rejects a file with no face, which fontdb otherwise accepts silently. - felis-client passes an empty list; its behavior is unchanged. - `docs/explanation/rendering/text-shaping.md`: "Font fallback" and "Font loading" describe the pinned path. A prebuilt `FontStack` handed to the constructor (the issue's proposal 3) was rejected: `start_fonts`, `reload_font` and the config's feature list could each reintroduce faces or features the stack was not built with. ## Verification - `just check` green. - New tests: discovery over the Monaspace Neon test files ignores a system fallback; a missing file and a non-font file each report `FontPath`; a warm-up stack is discarded when `font_files` changes. - pi review (sol): round 1 blocked the prebuilt-stack design; round 2 on this design passed with no findings.

natsukium added 1 commit 2026-09-26 00:31:18 +09:00

render, shaping: let a caller pin the font files a frame is drawn with ... e44360a92d

Font selection always scanned the system, so the same recording or
screenshot rendered with different faces on two machines, and a golden
test could not hold its pixels. A windowless consumer needs to choose
the exact font files.

`RendererConfig::font_files` replaces the system scan with the given
files for every lookup: the discovery `auto_discover` runs, styled faces
and fallback walk included, runs over them instead. It is part of the
config rather than a prebuilt stack handed to the constructor, so the
warm-up thread, the initial build and `reload_font` all resolve from the
same source, and the font features stay the config's as they do for a
window. A single-face `FontStack::new` was not enough: it draws bold,
italic and fallback glyphs differently from a window using the same
fonts.

The files are read into memory instead of mapped. The mmap path relies
on system font roots being immutable, and a caller's file can be
truncated under the mapping, which faults the process. A file with no
face in it is an error: fontdb accepts one silently, which would let a
typo fall through to a different face.

Assisted-by: Claude Code

natsukium merged commit a3d420ae01 into main 2026-09-26 00:45:07 +09:00

natsukium deleted branch render-pinned-fonts 2026-09-26 00:45:07 +09:00

natsukium referenced this pull request from a commit 2026-09-26 00:45:08 +09:00

render, shaping: let a caller pin the font files a frame is drawn with (#466)

natsukium referenced this pull request from natsukium/felis-reel 2026-09-26 00:50:51 +09:00

cli: pin the fonts a recording is drawn with (--font-file) #4

Commenting is not possible because the repository is archived.

Reviewers

No reviewers

Labels

Clear labels   

priority/P0

Must be resolved before the v0.1.0 tag

  

priority/P1

Must be settled before the first public contract freezes

  

priority/P2

Explicitly deferred until after v0.1.0

  

release/v0.1.0

Work tracked for the first tagged release

  

status/blocked

Cannot start now: waits on an upstream release, another issue, or a recorded revisit trigger

  

status/planned

Triaged with a source-grounded plan and a passed pi review; ready to implement

  

type/bug

Shipped behavior that contradicts the docs or the code's own intent; needs a product code change

  

type/design

An open scope or trade-off question; needs a recorded decision in the owning explanation doc before code

  

type/test-gap

Documented or reviewed behavior that no test pins; usually split from a review finding, needs no product code change

  

type/tracker

Parent issue that tracks child work

No labels

priority/P0

priority/P1

priority/P2

release/v0.1.0

status/blocked

status/planned

type/bug

type/design

type/test-gap

type/tracker

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

natsukium/felis!466

No description provided.