Port resvg to `harfrust` and `fontations` by nicoburns · Pull Request #922 · linebender/resvg

nicoburns · 2025-06-11T00:08:07Z

Changes made

Replaces rustybuzz with harfrust
Replaces ttf_parser with skrifa

Notes

This is failing 23 tests. These probably need to be manually checked. I'm waiting on Improve the test harness #954 to investigate these.
This requires an MSRV bump from 1.65 to 1.85
This will represent a binary size bump for users of resvg.

Production profile

The analysis below uses the following "production" profile:

[profile.production]
inherits = "release"
opt-level = 3
debug = false
lto = true
codegen-units = 1
strip = true
incremental = false

Binary size analysis:

According to cargo-bloat (not the most accurate):

harfrust and rustybuzz are both ~250kb
ttf_parser and read-fonts are both ~100kb
However, ttf_parser includes the metrics outlining functionality of skrifa. And skrifa is another ~250kb.

Compiling the CLI (which is a much more useful test) with the production profile above (cargo build --profile production --bin resvg) I get:

2.4mb for main
3.0mb for this PR

So this PR causes a binary size increase of around 600kb

Compile time analysis

Compile times on an M1 Pro (8+2 cores) are shown below (cargo build):

Profile	`main`	This PR
`debug`	4.68s	9.09s
`release`	7.64s	14.84s
`production`	24.59s	34.00s

So this PR roughly doubles compile times at regular opt settings
(with the effect being less pronounced with LTO and other production optimisation settings enabled)

LaurenzV · 2025-06-11T07:19:45Z

Yeah I think we first have to figure out how to deal with fontdb before doing anything with resvg. I don’t think RazrFalcon would want to switch it, and I’m not sure that fontique supports everything we need?

RazrFalcon · 2025-06-11T07:53:00Z

It's an interesting experiment, but I don't see a point in it. Is there a reason in switching to harfrust?
Last time I have checked it was still a beta, while rustybuzz (with all its faults) is a pretty robust solution.

PS: even if harfrust is faster, it doesn't matter either, because shaping isn't a bottleneck, lack of glyphs caching is.

RazrFalcon · 2025-06-11T07:54:50Z

Oh wait, it depends on proc-macros/syn?! Burn it with fire! No proc-macros in resvg.

LaurenzV · 2025-06-11T08:37:27Z

Last time I have checked it was still a beta, while rustybuzz (with all its faults) is a pretty robust solution.

harfrust is rustybuzz, just with ttf-parser switched out by read-fonts. It passes all the same tests, so it should be no less robust than then rustybuzz.

Is there a reason in switching to harfrust?

The reason is that going forward, (most likely) no one will be continually developing ttf-parser/rustybuzz, while harfrust will be actively maintained.

RazrFalcon · 2025-06-11T17:17:56Z

I would suggest to come back to it in a year. No rush.

And once again, proc-macros are no go. We have to figure out that one as well.

xStrom · 2025-06-12T16:25:46Z

Last time I have checked it was still a beta

Not sure about 'beta' type labels, but the first release (0.1.0) of HarfRust happened on June 10, marking a significant milestone.

It's an interesting experiment, but I don't see a point in it. Is there a reason in switching to harfrust?

For an immediate effect, it seems to be far more compliant when tested against the HarfBuzz test suite.

To quote Chad:

We’re still mulling over some public API changes and need to do some serious performance work but this is fully backed by fontations, up to date with HarfBuzz 11.2.1, and passes a significantly larger portion of the HB test suite than RustyBuzz: latest CI shows 5911 tests passing for HarfRust vs 2267 for RustyBuzz.

And the HarfRust introduction post by Behdad contains the following claim:

At the time of this writing, HarfRust passes most of the HarfBuzz shaping test suites, failing just 23 tests out of more than 6,000.

Moving forward, HarfRust will receive active development and continuous sync with HarfBuzz, while RustyBuzz is likely to only see critical fixes.

RazrFalcon · 2025-06-12T16:31:11Z

latest CI shows 5911 tests passing for HarfRust vs 2267 for RustyBuzz.

Apples and oranges, but sure.

Either way, proc-macros is the current major roadblock.

mattfbacon · 2025-06-17T13:38:54Z

@RazrFalcon Seems like the proc-macros are coming from font-types which uses bytemuck and serde derive macros. Do we want to talk to them about converting to manual implementations?

RazrFalcon · 2025-06-17T16:43:03Z

@mattfbacon if it's an easy fix - sure.

mattfbacon · 2025-06-17T19:45:07Z

No it wouldn't be at all, since the bytemuck derive macros generate unsafe code so they would need to allow unsafe code back into the codebase.

But what is your plan here? Earlier you said wait, but for what?

RazrFalcon · 2025-06-18T07:34:53Z

No it wouldn't be at all, since the bytemuck derive macros generate unsafe code so they would need to allow unsafe code back into the codebase.

I would call it easy. It's not like the code deeply depends on proc-macros.

But what is your plan here? Earlier you said wait, but for what?

Adoption, etc. I don't see a point/reason in switching to 0.1. Especially if we would keep ttf-parser around. It just doesn't make any sense.

Shnatsel · 2025-07-29T08:32:17Z

There are derive helpers based on macro_rules! rather than proc macros: https://matx.com/research/rules_derive

It is likely possible to implement an alternative derive for bytemuck traits using such helpers. These can then be published as a separate crate such as bytemuck_derive_lite to avoid any potential backward compatibility issues with the proc macro implementation.

xStrom · 2025-09-04T17:41:46Z

Is there a reason in switching to harfrust?

As some months have passed another benefit has appeared - performance. In the last month the HarfRust vs HarfBuzz OpenType shaping performance difference has gone from ~4x to ~1.2x, which is a massive boost. Not quite on par with HarfBuzz yet, but getting a lot closer. More importantly for this PR here though, HarfRust is now a lot faster than RustyBuzz.

Enivex · 2025-09-10T15:52:26Z

The geometric mean of the last result is ~1.16, meaning Harfrust is now only 16% slower than Harfbuzz.

behdad · 2025-09-10T15:54:28Z

The geometric mean of the last result is ~1.16, meaning Harfrust is now only 16% slower than Harfbuzz.

It's even better than that. See the comment:
harfbuzz/harfrust#257 (comment)

In short, there's some hb-harfrust overhead included in the spreadsheet number.

Skrifa's Matrix multiplication chains right-to-left (C = A * B applies B then A), so nested paint transforms must be pre-concatenated onto the current transform, matching the previous ttf-parser Transform::combine(self.transform, transform) behaviour and skrifa's own CollectFillGlyphPainter.

The single-glyph vs multi-glyph document heuristic was checking the first record in the SVG table instead of the record that actually contains the requested glyph, picking the wrong branch for fonts whose records differ (e.g. Noto Color Emoji).

Replace unwraps and direct slice indexing with graceful fallbacks: - palette lookups fall back to the foreground color instead of panicking on a missing/malformed CPAL table or an out-of-range palette index - clip glyph outlining returns early on a draw error, matching the old ttf-parser behaviour

CPAL color records for all palettes live in a single shared array; palette entry indices must be offset by the start index of the palette from color_record_indices rather than indexing the array directly. This is usually 0 for the first palette, but not guaranteed to be.

Signed-off-by: Nico Burns <nico@nicoburns.com>

Make font metrics variation-aware Resolve the requested font-variation-settings to a normalized location in load_font so that skrifa applies MVAR deltas to ascent, descent, x-height, underline and strikeout metrics, and so the manual subscript/superscript (SBYO/SPYO) delta handling actually runs. Metrics are cached per Font spec (which includes variations), so caching remains correct. Also drop the leftover commented-out ttf-parser code in outline(): with skrifa, the default (empty) location already resolves to the default value of every axis. Cache variable font outlines per variation set Previously any span with explicit variations (or auto optical sizing on a font with an opsz axis) bypassed the glyph outline cache entirely, re-extracting the outline for every glyph occurrence. Instead, compute the effective variation list (explicit settings plus auto opsz) once per glyph via a shared helper, include it in the cache key, and merge outline_with_variations into a single cached outline function. An empty variation list resolves to the default location, so non-variable fonts behave as before. Apply font variations to COLR color glyphs COLR glyphs were always painted at the default variation location, ignoring font-variation-settings. Resolve the effective variations to a normalized location and pass it to skrifa paint(), which resolves all COLR ItemVariationStore deltas (paint transforms, gradient stops, clip boxes). The GlyphPainter now also draws clip glyph outlines at the same location so the base outlines vary too. Color glyph trees are cached per variation set, like outlines.

Skrifa pairs push_clip_glyph with pop_clip (including via the default fill_glyph decomposition), but push_clip_glyph never opened an XML element while pop_clip unconditionally closed one. Each painted layer therefore closed an ancestor element, truncating the generated SVG and making Tree::from_data fail for color glyphs with more than a few layers. This went unnoticed because the Noto COLR test font also contains an SVG table, so rendering silently fell back to the SVG glyph path. push_clip_glyph now always opens a clip group with the glyph outline (an empty path when outlining fails), keeping pop_clip balanced. The colrv0/colrv1 snapshots change because COLR rendering now takes priority over the SVG table again; the output matches main.

ColorPalettes handles per-palette record offsets internally, so the manual color_record_indices base-offset handling is no longer needed.

Replaces the manual fvar table iteration in has_opsz_axis and the shaping auto-opsz check, and drops the harfrust FontRef usage from flatten.rs (skrifa and harfrust share the same underlying types).

The instance only exists to apply variation settings, so avoid the setup cost for non-variable text (the common case).

ttf-parser enforced 16..=16384 when parsing the head table; skrifa does not, so add the check to load_font to avoid nonsensical metrics from malformed fonts.

skrifa reports metrics as f32; round when converting to the integer font units used by ResolvedFont rather than truncating toward zero, matching the integer values ttf-parser used to return. Also ignore non-positive x-height values, like ttf-parser did.

The painter previously reused the last outlined glyph (path_buf) as the fill geometry, which only works when exactly one clip glyph is active. The COLR imaging model instead has a proper clip stack: a fill paints the intersection of all currently pushed clips. - fill() now paints a rectangle covering the intersection of the tracked clip bounds (in root space), shaped by the enclosing SVG clip groups, instead of whatever happened to be in path_buf. - push_clip_glyph/push_clip_box track their bounds on a clip stack. - fill_glyph() is implemented directly (fill the glyph outline with the brush), avoiding the default clip-then-fill decomposition. This also restores the exact rendering of the old ttf-parser painter for COLRv0, so colrv0.png reverts to the version on main. The colrv1 snapshot changes slightly at glyph edges: leaf fills no longer pass through a clip of the same shape, so edge anti-aliasing is no longer applied twice.

Co-authored-by: Laurenz Stampfl <47084093+LaurenzV@users.noreply.github.com>

Signed-off-by: Nico Burns <nico@nicoburns.com>

nicoburns · 2026-07-29T18:35:28Z

I've published fontdb v0.24 (which inlines a minimal subset of ttf-parser) and updated this PR to use the version from crates.io.

Signed-off-by: Nico Burns <nico@nicoburns.com>

nicoburns · 2026-08-02T12:13:02Z

This has now been published as part of the v0.48 release

@reed-smout

##### [v0.48.0](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0480-2026-07-31) This release has an MSRV of 1.89.0 for `usvg` and `resvg` and the C API. The big change in this release is that text support is now backed by [`skrifa`](https://github.com/googlefonts/fontations) and [`harfrust`](https://github.com/harfbuzz/harfrust) rather than [`ttf-parser`](https://github.com/harfbuzz/ttf-parser) and [`rustybuzz`](https://github.com/harfbuzz/rustybuzz): - Moves resvg onto an actively maintained font stack. - Should have faster and more correct text rendering. - Will enable further improvements to things like variable font rendering in future. - May impact binary size (expected +\~500kb for people not otherwise including fontations dependencies in their tree). - May result in small rendering changes compared to older versions of resvg. ##### Added - New `svgz` and `writer` feature gates to reduce the number of required dependencies. ([#1088](linebender/resvg#1088) by [@reed-smout](https://github.com/reed-smout)) - Warnings when parsing malformed paths. ([#1011](linebender/resvg#1011)) - A `Display` implementation for `Units`. ([#986](linebender/resvg#986)) ##### Changed - MSRV bumped from 1.87 to 1.89. The requirement comes from `font-types`. - Text shaping now uses `harfrust` instead of `rustybuzz`, and font parsing uses `skrifa` (fontations) instead of `ttf-parser`. ([#922](linebender/resvg#922) by [@nicoburns](https://github.com/nicoburns)) - Filter input dimension checks are now debug-only assertions. ([#991](linebender/resvg#991)) ##### Fixed - Glyph advances are now calculated correctly. ([#1043](linebender/resvg#1043) by [@fundon](https://github.com/fundon)) - Text nodes now inherit their absolute transform. ([#1040](linebender/resvg#1040) by [@fundon](https://github.com/fundon)) - Fixes related to non-finite values. ([#1049](linebender/resvg#1049) by [@SAY-5](https://github.com/SAY-5)) - Transforms are no longer applied twice in `abs_transform`. ([#1056](linebender/resvg#1056) by [@fundon](https://github.com/fundon)) - `fr` is now resolved for radial gradients referenced via `href`. ([#1098](linebender/resvg#1098) by [@T1mVo](https://github.com/T1mVo)) - Panic in `feComposite` with the `arithmetic` operator when the filter region is larger than the clamped layer. ([#1021](linebender/resvg#1021), [#1007](linebender/resvg#1007)) - Application of `transform` (and other group properties like `opacity`) on a nested `svg` element. - The missing dimension of an `svg` with only `width` or `height` specified is now computed from its `viewBox` aspect ratio. ([#1045](linebender/resvg#1045)) - The `wght` variation coordinate is now set even when `font-weight` has its default value. ([#1099](linebender/resvg#1099)) - The unprefixed `href` attribute now takes precedence over the deprecated `xlink:href` when both are present, as required by SVG 2. ([#1015](linebender/resvg#1015)) - Incorrect y-axis offsets when transforming `feSpotLight` sources. ([#1052](linebender/resvg#1052)) - Panics caused by bounding boxes that exceed the supported integer range. ([#989](linebender/resvg#989)) ##### [v0.47.0](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0470-2026-02-05) This release has an MSRV of 1.87.0 for `usvg` and `resvg` and the C API. ##### Added - Focal radius (`fr`) supported for Radial Gradients. ([#1014](linebender/resvg#1014) by [@wmedrano](https://github.com/wmedrano)) - Support for variable fonts based on font-variation-settings CSS property. ([#997](linebender/resvg#997) by [@oetiker](https://github.com/oetiker)) ##### Changed - `tiny-skia` has a major version bump from 0.11 to 0.12.

@reed-smout

##### [v0.48.0](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0480-2026-07-31) This release has an MSRV of 1.89.0 for `usvg` and `resvg` and the C API. The big change in this release is that text support is now backed by [`skrifa`](https://github.com/googlefonts/fontations) and [`harfrust`](https://github.com/harfbuzz/harfrust) rather than [`ttf-parser`](https://github.com/harfbuzz/ttf-parser) and [`rustybuzz`](https://github.com/harfbuzz/rustybuzz): - Moves resvg onto an actively maintained font stack. - Should have faster and more correct text rendering. - Will enable further improvements to things like variable font rendering in future. - May impact binary size (expected +\~500kb for people not otherwise including fontations dependencies in their tree). - May result in small rendering changes compared to older versions of resvg. ##### Added - New `svgz` and `writer` feature gates to reduce the number of required dependencies. ([#1088](linebender/resvg#1088) by [@reed-smout](https://github.com/reed-smout)) - Warnings when parsing malformed paths. ([#1011](linebender/resvg#1011)) - A `Display` implementation for `Units`. ([#986](linebender/resvg#986)) ##### Changed - MSRV bumped from 1.87 to 1.89. The requirement comes from `font-types`. - Text shaping now uses `harfrust` instead of `rustybuzz`, and font parsing uses `skrifa` (fontations) instead of `ttf-parser`. ([#922](linebender/resvg#922) by [@nicoburns](https://github.com/nicoburns)) - Filter input dimension checks are now debug-only assertions. ([#991](linebender/resvg#991)) ##### Fixed - Glyph advances are now calculated correctly. ([#1043](linebender/resvg#1043) by [@fundon](https://github.com/fundon)) - Text nodes now inherit their absolute transform. ([#1040](linebender/resvg#1040) by [@fundon](https://github.com/fundon)) - Fixes related to non-finite values. ([#1049](linebender/resvg#1049) by [@SAY-5](https://github.com/SAY-5)) - Transforms are no longer applied twice in `abs_transform`. ([#1056](linebender/resvg#1056) by [@fundon](https://github.com/fundon)) - `fr` is now resolved for radial gradients referenced via `href`. ([#1098](linebender/resvg#1098) by [@T1mVo](https://github.com/T1mVo)) - Panic in `feComposite` with the `arithmetic` operator when the filter region is larger than the clamped layer. ([#1021](linebender/resvg#1021), [#1007](linebender/resvg#1007)) - Application of `transform` (and other group properties like `opacity`) on a nested `svg` element. - The missing dimension of an `svg` with only `width` or `height` specified is now computed from its `viewBox` aspect ratio. ([#1045](linebender/resvg#1045)) - The `wght` variation coordinate is now set even when `font-weight` has its default value. ([#1099](linebender/resvg#1099)) - The unprefixed `href` attribute now takes precedence over the deprecated `xlink:href` when both are present, as required by SVG 2. ([#1015](linebender/resvg#1015)) - Incorrect y-axis offsets when transforming `feSpotLight` sources. ([#1052](linebender/resvg#1052)) - Panics caused by bounding boxes that exceed the supported integer range. ([#989](linebender/resvg#989)) ##### [v0.47.0](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0470-2026-02-05) This release has an MSRV of 1.87.0 for `usvg` and `resvg` and the C API. ##### Added - Focal radius (`fr`) supported for Radial Gradients. ([#1014](linebender/resvg#1014) by [@wmedrano](https://github.com/wmedrano)) - Support for variable fonts based on font-variation-settings CSS property. ([#997](linebender/resvg#997) by [@oetiker](https://github.com/oetiker)) ##### Changed - `tiny-skia` has a major version bump from 0.11 to 0.12.

@reed-smout

##### [v0.48.1](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0481-2026-08-02) This release has an MSRV of 1.85.0 for `usvg` and `resvg` and the C API. ##### Changed - MSRV lowered from 1.89 to 1.85. ([#1109](linebender/resvg#1109)) - Downgraded strict-num dependency to 0.1.1. This prevents duplicate versions from being included ([#1110](linebender/resvg#1110)) ##### Fixed - Fixed an `as_mut_slice` future-compatibility warning. ([#1108](linebender/resvg#1108)) ##### [v0.48.0](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0480-2026-07-31) This release has an MSRV of 1.89.0 for `usvg` and `resvg` and the C API. The big change in this release is that text support is now backed by [`skrifa`](https://github.com/googlefonts/fontations) and [`harfrust`](https://github.com/harfbuzz/harfrust) rather than [`ttf-parser`](https://github.com/harfbuzz/ttf-parser) and [`rustybuzz`](https://github.com/harfbuzz/rustybuzz): - Moves resvg onto an actively maintained font stack. - Should have faster and more correct text rendering. - Will enable further improvements to things like variable font rendering in future. - May impact binary size (expected +\~500kb for people not otherwise including fontations dependencies in their tree). - May result in small rendering changes compared to older versions of resvg. ##### Added - New `svgz` and `writer` feature gates to reduce the number of required dependencies. ([#1088](linebender/resvg#1088) by [@reed-smout](https://github.com/reed-smout)) - Warnings when parsing malformed paths. ([#1011](linebender/resvg#1011)) - A `Display` implementation for `Units`. ([#986](linebender/resvg#986)) ##### Changed - MSRV bumped from 1.87 to 1.89. The requirement comes from `font-types`. - Text shaping now uses `harfrust` instead of `rustybuzz`, and font parsing uses `skrifa` (fontations) instead of `ttf-parser`. ([#922](linebender/resvg#922) by [@nicoburns](https://github.com/nicoburns)) - Filter input dimension checks are now debug-only assertions. ([#991](linebender/resvg#991)) ##### Fixed - Glyph advances are now calculated correctly. ([#1043](linebender/resvg#1043) by [@fundon](https://github.com/fundon)) - Text nodes now inherit their absolute transform. ([#1040](linebender/resvg#1040) by [@fundon](https://github.com/fundon)) - Fixes related to non-finite values. ([#1049](linebender/resvg#1049) by [@SAY-5](https://github.com/SAY-5)) - Transforms are no longer applied twice in `abs_transform`. ([#1056](linebender/resvg#1056) by [@fundon](https://github.com/fundon)) - `fr` is now resolved for radial gradients referenced via `href`. ([#1098](linebender/resvg#1098) by [@T1mVo](https://github.com/T1mVo)) - Panic in `feComposite` with the `arithmetic` operator when the filter region is larger than the clamped layer. ([#1021](linebender/resvg#1021), [#1007](linebender/resvg#1007)) - Application of `transform` (and other group properties like `opacity`) on a nested `svg` element. - The missing dimension of an `svg` with only `width` or `height` specified is now computed from its `viewBox` aspect ratio. ([#1045](linebender/resvg#1045)) - The `wght` variation coordinate is now set even when `font-weight` has its default value. ([#1099](linebender/resvg#1099)) - The unprefixed `href` attribute now takes precedence over the deprecated `xlink:href` when both are present, as required by SVG 2. ([#1015](linebender/resvg#1015)) - Incorrect y-axis offsets when transforming `feSpotLight` sources. ([#1052](linebender/resvg#1052)) - Panics caused by bounding boxes that exceed the supported integer range. ([#989](linebender/resvg#989)) ##### [v0.47.0](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0470-2026-02-05) This release has an MSRV of 1.87.0 for `usvg` and `resvg` and the C API. ##### Added - Focal radius (`fr`) supported for Radial Gradients. ([#1014](linebender/resvg#1014) by [@wmedrano](https://github.com/wmedrano)) - Support for variable fonts based on font-variation-settings CSS property. ([#997](linebender/resvg#997) by [@oetiker](https://github.com/oetiker)) ##### Changed - `tiny-skia` has a major version bump from 0.11 to 0.12.

@reed-smout

##### [v0.48.1](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0481-2026-08-02) This release has an MSRV of 1.85.0 for `usvg` and `resvg` and the C API. ##### Changed - MSRV lowered from 1.89 to 1.85. ([#1109](linebender/resvg#1109)) - Downgraded strict-num dependency to 0.1.1. This prevents duplicate versions from being included ([#1110](linebender/resvg#1110)) ##### Fixed - Fixed an `as_mut_slice` future-compatibility warning. ([#1108](linebender/resvg#1108)) ##### [v0.48.0](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0480-2026-07-31) This release has an MSRV of 1.89.0 for `usvg` and `resvg` and the C API. The big change in this release is that text support is now backed by [`skrifa`](https://github.com/googlefonts/fontations) and [`harfrust`](https://github.com/harfbuzz/harfrust) rather than [`ttf-parser`](https://github.com/harfbuzz/ttf-parser) and [`rustybuzz`](https://github.com/harfbuzz/rustybuzz): - Moves resvg onto an actively maintained font stack. - Should have faster and more correct text rendering. - Will enable further improvements to things like variable font rendering in future. - May impact binary size (expected +\~500kb for people not otherwise including fontations dependencies in their tree). - May result in small rendering changes compared to older versions of resvg. ##### Added - New `svgz` and `writer` feature gates to reduce the number of required dependencies. ([#1088](linebender/resvg#1088) by [@reed-smout](https://github.com/reed-smout)) - Warnings when parsing malformed paths. ([#1011](linebender/resvg#1011)) - A `Display` implementation for `Units`. ([#986](linebender/resvg#986)) ##### Changed - MSRV bumped from 1.87 to 1.89. The requirement comes from `font-types`. - Text shaping now uses `harfrust` instead of `rustybuzz`, and font parsing uses `skrifa` (fontations) instead of `ttf-parser`. ([#922](linebender/resvg#922) by [@nicoburns](https://github.com/nicoburns)) - Filter input dimension checks are now debug-only assertions. ([#991](linebender/resvg#991)) ##### Fixed - Glyph advances are now calculated correctly. ([#1043](linebender/resvg#1043) by [@fundon](https://github.com/fundon)) - Text nodes now inherit their absolute transform. ([#1040](linebender/resvg#1040) by [@fundon](https://github.com/fundon)) - Fixes related to non-finite values. ([#1049](linebender/resvg#1049) by [@SAY-5](https://github.com/SAY-5)) - Transforms are no longer applied twice in `abs_transform`. ([#1056](linebender/resvg#1056) by [@fundon](https://github.com/fundon)) - `fr` is now resolved for radial gradients referenced via `href`. ([#1098](linebender/resvg#1098) by [@T1mVo](https://github.com/T1mVo)) - Panic in `feComposite` with the `arithmetic` operator when the filter region is larger than the clamped layer. ([#1021](linebender/resvg#1021), [#1007](linebender/resvg#1007)) - Application of `transform` (and other group properties like `opacity`) on a nested `svg` element. - The missing dimension of an `svg` with only `width` or `height` specified is now computed from its `viewBox` aspect ratio. ([#1045](linebender/resvg#1045)) - The `wght` variation coordinate is now set even when `font-weight` has its default value. ([#1099](linebender/resvg#1099)) - The unprefixed `href` attribute now takes precedence over the deprecated `xlink:href` when both are present, as required by SVG 2. ([#1015](linebender/resvg#1015)) - Incorrect y-axis offsets when transforming `feSpotLight` sources. ([#1052](linebender/resvg#1052)) - Panics caused by bounding boxes that exceed the supported integer range. ([#989](linebender/resvg#989)) ##### [v0.47.0](https://github.com/linebender/resvg/blob/HEAD/CHANGELOG.md#0470-2026-02-05) This release has an MSRV of 1.87.0 for `usvg` and `resvg` and the C API. ##### Added - Focal radius (`fr`) supported for Radial Gradients. ([#1014](linebender/resvg#1014) by [@wmedrano](https://github.com/wmedrano)) - Support for variable fonts based on font-variation-settings CSS property. ([#997](linebender/resvg#997) by [@oetiker](https://github.com/oetiker)) ##### Changed - `tiny-skia` has a major version bump from 0.11 to 0.12.

nicoburns force-pushed the harfrust branch 2 times, most recently from 02f55d3 to 62c028b Compare June 11, 2025 01:04

xStrom reviewed Jun 12, 2025

View reviewed changes

Comment thread .github/workflows/main.yml Outdated

nicoburns force-pushed the harfrust branch from 62c028b to 5b0e481 Compare August 14, 2025 23:34

nicoburns force-pushed the harfrust branch 4 times, most recently from e29cfbe to 0973b97 Compare August 31, 2025 11:18

nicoburns changed the title ~~Port resvg to harfrust~~ Port resvg to harfrust and fontations Aug 31, 2025

nicoburns force-pushed the harfrust branch 3 times, most recently from 3c5c998 to c05e2fd Compare August 31, 2025 19:20

nicoburns mentioned this pull request Aug 31, 2025

Tracking: use Rust font libraries (HarfRust, Skrifa/Fontations, etc) servo/servo#33482

Open

4 tasks

nicoburns force-pushed the harfrust branch 2 times, most recently from da2754d to 7e549e1 Compare September 13, 2025 12:17

nicoburns and others added 18 commits July 29, 2026 11:30

Add pop_layer function

a7d9bf0

Signed-off-by: Nico Burns <nico@nicoburns.com>

Hoist shaper construction out of the loop layout.rs

e463481

Signed-off-by: Nico Burns <nico@nicoburns.com>

Optimize new snapshots with oxipng

c237a4b

Signed-off-by: Nico Burns <nico@nicoburns.com>

Use skrifa ColorPalettes API for CPAL lookups

eb6de1d

ColorPalettes handles per-palette record offsets internally, so the manual color_record_indices base-offset handling is no longer needed.

Use axes().get_by_tag() for opsz axis checks

9fca217

Replaces the manual fvar table iteration in has_opsz_axis and the shaping auto-opsz check, and drops the harfrust FontRef usage from flatten.rs (skrifa and harfrust share the same underlying types).

Skip building a ShaperInstance when there are no variations

ac32b46

The instance only exists to apply variation settings, so avoid the setup cost for non-variable text (the common case).

Reject fonts with an out-of-range unitsPerEm

1580c77

ttf-parser enforced 16..=16384 when parsing the head table; skrifa does not, so add the check to load_font to avoid nonsensical metrics from malformed fonts.

Deduplicate OPSZ constant

9f8a32c

Use u32 for GlyphId

5722ea3

Remove COLR comment

3c2026b

Co-authored-by: Laurenz Stampfl <47084093+LaurenzV@users.noreply.github.com>

nicoburns force-pushed the harfrust branch from 59d6873 to 3c2026b Compare July 29, 2026 18:30

Use fontdb 0.24

0507958

Signed-off-by: Nico Burns <nico@nicoburns.com>

nicoburns added 2 commits July 29, 2026 14:53

Remove println's from colr module

21ed498

Update TODO comments

a99aea4

Signed-off-by: Nico Burns <nico@nicoburns.com>

nicoburns merged commit 4d76f92 into main Jul 29, 2026
10 checks passed

Its-Just-Nans mentioned this pull request Jul 30, 2026

Clamp too large bboxes instead of panicking #989

Merged

Uh oh!

Conversation

nicoburns commented Jun 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes made

Notes

Production profile

Binary size analysis:

Compile time analysis

Uh oh!

LaurenzV commented Jun 11, 2025

Uh oh!

RazrFalcon commented Jun 11, 2025

Uh oh!

RazrFalcon commented Jun 11, 2025

Uh oh!

LaurenzV commented Jun 11, 2025

Uh oh!

RazrFalcon commented Jun 11, 2025

Uh oh!

Uh oh!

xStrom commented Jun 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

RazrFalcon commented Jun 12, 2025

Uh oh!

mattfbacon commented Jun 17, 2025

Uh oh!

RazrFalcon commented Jun 17, 2025

Uh oh!

mattfbacon commented Jun 17, 2025

Uh oh!

RazrFalcon commented Jun 18, 2025

Uh oh!

Shnatsel commented Jul 29, 2025

Uh oh!

xStrom commented Sep 4, 2025

Uh oh!

Enivex commented Sep 10, 2025

Uh oh!

behdad commented Sep 10, 2025

Uh oh!

nicoburns commented Jul 29, 2026

Uh oh!

Uh oh!

nicoburns commented Aug 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

12 participants

nicoburns commented Jun 11, 2025 •

edited

Loading

xStrom commented Jun 12, 2025 •

edited

Loading