|
c124252d
|
2025-05-31T16:32:58
|
|
[algs] Add support for little-endian int types
As well as 64bit-int, and float types.
Add Align type.
|
|
|
47089fe7
|
2025-05-28T17:10:22
|
|
Big misc changes
- Make geometry types into templates taking float or double.
- Changes to paint() transform elimination.
- Add font->has_nonzero_coords and use it across.
- Extensions to avar processing for otherwise invalid mappings.
|
|
|
e1dcc454
|
2025-05-21T16:00:08
|
|
[gvar] Use cached scratch in get_extents
Makes it alloc-free.
15% faster on benchmark-font glyph_extents RobotoFlex var benchmark.
|
|
|
c6097698
|
2025-05-16T01:10:43
|
|
[ot-font/draw] Use a gvar shared-tuple scalar cache for calculate_scalar
Massive saving on massively variable fonts. Over 30% on a tested font.
|
|
|
f50490d6
|
2025-05-15T18:28:44
|
|
Add hb_scalar_cache_t
|
|
|
89e28a23
|
2025-05-10T16:30:35
|
|
[skippy-iter] Use auto for type deduction
|
|
|
59334682
|
2025-04-23T20:56:32
|
|
[shape] Only apply slant to shaping results if horizontal
|
|
|
e71a88d3
|
2025-04-23T19:17:31
|
|
[paint] Fix crash
|
|
|
6688a03c
|
2025-04-23T19:06:52
|
|
[sbix/CBDT] Fix memory leaks in error case
|
|
|
17238eae
|
2025-04-23T18:15:56
|
|
[ot/ft/fontations] Port to new _or_fail of draw/paint APIs
|
|
|
a23021e1
|
2025-04-23T00:43:56
|
|
[COLRv1] Simplify paint_glyph()
Pass a non-synthetic font to the client, so they don't have
to worry about it.
The only issue left, is that now when paint_glyph() paints
an outline glyph (not color glyph), it would be double-slanted.
I'm leaning towards removing painting of outline glyphs from
the paint API and change the paint API to return a boolean.
|
|
|
01e9c77e
|
2025-04-22T21:57:02
|
|
[COLR] Fix clip-box
Was slanting it. Was wrong with new code.
|
|
|
a69713b5
|
2025-04-22T21:01:04
|
|
[color] Move slanting to font layer
COLRv1 is still being double-slanted. Investigating.
|
|
|
05362548
|
2025-04-20T05:06:54
|
|
Merge pull request #5294 from harfbuzz/paint-bounds
[COLRv1] `paint-bounds`
|
|
|
27b2e16a
|
2025-04-20T04:34:40
|
|
[atomic] Make cmpexch() non-const
Was a mistake. Make the affected members mutable instead.
|
|
|
78822b63
|
2025-04-20T04:23:35
|
|
Revert "[COLRv1] Move paint boundedness check to font layer"
This reverts commit c9a26fbf47f01fdd55df1a0ccc81111cb07bf108.
|
|
|
c9a26fbf
|
2025-04-20T04:22:32
|
|
[COLRv1] Move paint boundedness check to font layer
This way it will apply to fontations and future implementations
as well.
But this slows down things apparently, specially if the font
does use clipboxes. So I'm going to revert.
|
|
|
8aa88aa4
|
2025-04-20T04:08:41
|
|
[COLRv1:ot/ft] Use paint-bounded instead of paint-extents
Faster. Note that this also means that if there is no clip-box,
we don't compute the bounds and emit an initial clip anymore.
That was totally unnecessary and against the COLRv1 rendering
algorithm.
This makes painting of COLRv1 fonts without clipboxes *much*
faster, as previously we were computing bounds using all subglyph
outlines. Not anymore.
Before:
BM_Font/paint_glyph/Nupuram-Color.colrv1.ttf/ot 1.78 ms
After:
BM_Font/paint_glyph/Nupuram-Color.colrv1.ttf/ot 0.155 ms
|
|
2602f729
|
2025-04-17T15:02:59
|
|
Avoid unannotated-fallthrough warning in COLR.hh
|
|
|
9cae040b
|
2025-04-03T22:18:14
|
|
[ItemVariationStore] Make cache atomic floats
Doesn't work with intrinsic atomic implementations. Disabled that.
Going through the C++11 atomics path now.
Will fix in next commit.
|
|
|
7f5fafec
|
2025-04-01T21:35:28
|
|
[GDEF] Simplify cache declaration
|
|
5ebe36e1
|
2025-03-27T13:09:18
|
|
[subset] bug fix for CaretValueFormat3
Don't return false in case of a Hinting Device table
|
|
|
192d264a
|
2025-03-25T13:32:01
|
|
[subset] remove divisions when computing the cost of binary search
|
|
|
14e5a046
|
2025-03-16T00:36:11
|
|
[font] Move synthetic glyph extents into font layer
So it applies to all font-funcs.
|
|
|
000a0ad7
|
2025-03-15T14:37:44
|
|
[paint] Add push_font_transform() / push_inverse_font_transform()
Fixes https://github.com/harfbuzz/harfbuzz/issues/5146
|
|
|
4807a021
|
2025-03-12T18:27:27
|
|
[atomic] Kill hb_atomic_ptr_t<T>
Use hb_atomic_t<T *> instead.
|
|
|
4e5358a2
|
2025-03-07T15:18:28
|
|
[COLRv1] Fix comments
Fixes https://github.com/harfbuzz/harfbuzz/issues/5116
|
|
|
575e70b0
|
2025-03-07T12:51:44
|
|
[COLR] Use palette 0 if palette index out of range
Fixes https://github.com/harfbuzz/harfbuzz/issues/5112
|
|
|
6506bd7e
|
2025-03-06T20:00:54
|
|
Merge pull request #4498 from harfbuzz/PaintColrLayers
[COLRv1] Optimize PaintColrLayers instead of PaintComposite
|
|
|
50fe45f4
|
2025-03-03T16:31:41
|
|
[COLR] Fix memory leak
|
|
|
9c6b6998
|
2025-03-03T11:46:22
|
|
[VARC] Fix sign of returned extents
|
|
|
e9134c8c
|
2025-03-03T02:04:34
|
|
[COLR] Reduce malloc pressure
By caching paint-extents.
|
|
|
9efd17e2
|
2025-03-03T01:44:01
|
|
[COLR] Switch to using (existing) accelerator
|
|
|
17ddc844
|
2025-03-03T01:27:32
|
|
[VARC] Fix scratch caching
|
|
|
a62058ee
|
2025-03-03T01:19:52
|
|
[VARC] Separate varc scratch from glyf scratch
|
|
|
3199d1c1
|
2025-03-02T18:35:46
|
|
[VARC] Use a context object to pass arguments around
|
|
|
e0c6c980
|
2025-03-02T13:32:54
|
|
[VARC] Implement get_extents()
Fixes https://github.com/harfbuzz/harfbuzz/issues/5007
|
|
|
7e3c9658
|
2025-03-01T15:08:41
|
|
[VARC] Reuse parent MultiVarStore cache if coords didn't change
Minor optimization; applicable to the test hangul font mostly.
|
|
|
64dbeaf0
|
2025-02-28T00:54:39
|
|
[glyf] Mover decycler to the scratch pad
|
|
|
00541f19
|
2025-02-26T16:22:48
|
|
Merge pull request #4970 from harfbuzz/GVAR
Add `GVAR` table
|
|
|
a0543453
|
2025-02-26T15:36:03
|
|
[CBDT] Change has_data() to be more accurate
get_length() was returning the Null table length if table
was not present.
|
|
|
7504c67f
|
2025-02-26T15:24:40
|
|
[CBDT] Return immediately if no data
|
|
|
17a83b74
|
2025-02-26T14:40:02
|
|
[COLRv1] Optimize PaintColrLayers instead of PaintComposite
See https://github.com/googlefonts/colr-gradients-spec/issues/369#issuecomment-1816913943
and onwards.
|
|
|
44b7ba51
|
2025-02-26T13:47:45
|
|
[GVAR] Hide behind HB_NO_BEYOND_64K
|
|
|
13900ce4
|
2025-02-26T13:41:52
|
|
[GVAR] Hook up to face and glyf table
|
|
|
6a7b4dca
|
2025-02-26T13:25:50
|
|
[VARC] Reduce stack usage on 32bit systems
|
|
|
84ffb7e4
|
2025-02-26T12:51:04
|
|
[SimpleGlyph] Minor use an intermediate variable
|
|
|
54db2b2e
|
2025-02-26T12:48:55
|
|
[SimpleGlyph] Remove a conditional from hot path
5% speedup in Roboto-Regular draw benchmark
|
|
|
bb4d37b0
|
2025-02-26T12:17:27
|
|
[VARC/coord-setter] Reduce stack allocation on 32bit systems
|
|
|
51d3489c
|
2025-02-26T12:09:31
|
|
[Composite] Fix memory-free rendering
|
|
|
59e6a1c4
|
2025-02-26T11:16:04
|
|
[glyf/Composite] Malloc-free
|
|
|
3c4cb472
|
2025-02-26T10:48:36
|
|
[VARC] Fix uninitialized value
Fixes https://oss-fuzz.com/testcase-detail/6089638800588800
|
|
|
fcbf14a0
|
2025-02-25T20:16:18
|
|
[CompositeGlyph] Adjust pre-allocation
|
|
|
6ca8852e
|
2025-02-25T19:57:43
|
|
[glyf] Return earlier if table is empty
|
|
|
7614320a
|
2025-02-25T16:59:37
|
|
[glyf] Reduce malloc pressure
Benchmark 15% faster for draw.
|
|
|
882d3288
|
2025-02-25T16:34:55
|
|
[glyf] Port to decycler for composite glyphs
5% speed up drawing Roboto-Regular.ttf.
|
|
|
ffee85d6
|
2025-02-25T16:20:42
|
|
[VARC] Minor simplify pointer math
|
|
|
7222c1e5
|
2025-02-25T15:13:20
|
|
[path-builder] Add a constexpr
|
|
|
e450552d
|
2025-02-25T14:54:37
|
|
[coord-setter] Fix memory access in case of malloc failure
Fixes https://oss-fuzz.com/testcase-detail/5383702943432704
|
|
|
f71faf5f
|
2025-02-25T11:35:21
|
|
[VARC] Fix buffer scratch borrowing
|
|
|
e76b689f
|
2025-02-25T11:23:48
|
|
[VARC] Fix resource leak
|
|
|
47124e63
|
2025-02-24T23:25:02
|
|
[VARC] Avoid creating cache if there's no VARC table
|
|
|
d5d199fb
|
2025-02-24T23:20:12
|
|
[VARC] Reduce malloc pressure
By caching one hb_glyf_scratch_t on the VARC accelerator.
|
|
|
2c5ab14a
|
2025-02-24T22:50:42
|
|
[coord-setter] Increase static array size
|
|
|
97a5c52a
|
2025-02-24T20:53:19
|
|
[VARC/MultiVarStore] Reduce malloc pressure more
|
|
|
6938ee03
|
2025-02-24T20:40:26
|
|
[VARC] More nuanced caching
|
|
|
081fcbdf
|
2025-02-24T20:15:24
|
|
[VARC] Minor move variable closer to use
|
|
|
0a1b26b8
|
2025-02-24T20:07:59
|
|
[coord-setter] Reduce malloc pressure
~15% speedup benchmark-font draw of varc-hanzi.ttf
|
|
|
76c3beaf
|
2025-02-24T19:50:51
|
|
[VARC] Reduce malloc overhead
|
|
|
2032da01
|
2025-02-18T20:54:56
|
|
[ot-font] Centralize painting an outline glyph
|
|
|
01f02f55
|
2025-02-18T20:28:48
|
|
[COLR] Optimize NoVariation codepath
|
|
|
4c263ecd
|
2025-02-18T20:08:19
|
|
Merge pull request #5074 from harfbuzz/tortoise-hare
[decycler] Implement an efficient graph cycle detector
|
|
|
f26d9514
|
2025-02-18T20:07:25
|
|
[COLR] Fix HB_LEAN build
|
|
|
0667ceae
|
2025-02-16T10:37:21
|
|
[VARC] Reduce stack use
|
|
|
4335e49a
|
2025-02-16T10:33:49
|
|
[VARC] Port to hb-decycler-t
5% faster on varc-hanzi test case.
|
|
|
0aa400b1
|
2025-02-15T23:19:44
|
|
[decycler] Implement an efficient graph cycle detector
This is an algorithm I came up with, based on the Floyd's
Tortoise-Hare constant-memory linear-time linked-list cycle-detection
algorithm.
https://en.wikipedia.org/wiki/Cycle_detection#Floyd's_tortoise_and_hare
It is linear-time and malloc-free. It *eventually* detects cycles,
not immediately.
The main different with Floyd's algorithm is that this algorithm
detects cycles when one is traversing down a graph, not just a
linked list.
Our existing cycle-detection algorithms use a set-of-integers,
either hb_set_t, or more efficient in this case, hb_map_t. Those
include at least one malloc, and as such show up on profiles.
Port hb-ot-font COLRv1 to use the decycler instead of previous
hb_map_t usage for cycle detection.
benchmark-font paint_glyph on NotoColorEmoji-Regular.ttf:
Before: 8ms; After: 5.5ms. No cycle detection: 5.5ms.
FT COLRv1 API is so slow (174ms) it's not worth porting to this.
Other graphs (VARC, etc) to be ported.
Test and documentation to be added.
|
|
|
ed76c855
|
2025-02-18T20:01:36
|
|
[COLR] Optimize palette access
|
|
|
016e78b6
|
2025-02-18T19:49:48
|
|
[COLR] Micro-optimize
|
|
|
d2a88fbd
|
2025-02-12T21:48:22
|
|
[vector] Add alloc_exact()
|
|
|
4b54ee11
|
2025-02-09T12:38:58
|
|
[glyf] Change drawing algorithm to match FreeType / CoreText
|
|
|
f68193a0
|
2025-02-09T11:43:07
|
|
[glyf] Use direct access to points array
Not necessary faster. Prep'ing for next change.
|
|
|
b8327fe6
|
2025-02-06T20:01:12
|
|
Fix compiler errors
|
|
|
c7378294
|
2025-02-06T01:42:25
|
|
[OT] Remove likely() from coverage check
Not that it matters, but because of the set-digest, this path
is not very likely() anymore.
|
|
|
cad264b2
|
2025-02-06T00:56:16
|
|
Indent
|
|
|
69057e6e
|
2025-02-05T19:50:38
|
|
[Ligature] Cache coverage the same way as PairPos1
Another 7% down on Roboto-Regular.
|
|
|
7a746c32
|
2025-02-05T19:38:25
|
|
[Coverage] Fix cache cost reporting
|
|
|
5be62a9f
|
2025-02-05T19:30:01
|
|
[PairPosFormat1] Use a coverage cache as well
Speeds up Arial shaping by 3%.
|
|
|
f22943a2
|
2025-02-05T18:51:11
|
|
[PairPos] Cache coverage as well
Another 3% down in Roboto-Regular.
|
|
|
726af2e4
|
2025-02-05T18:10:40
|
|
[PairPos] Use a class cache
10% speedup on Roboto-Regular. 5% on SF Pro.
|
|
|
4df11621
|
2025-01-24T05:00:05
|
|
[VARC] Fix variable binding
Ouch.
|
|
|
2fa9ccee
|
2025-01-23T23:13:49
|
|
[VARC] Simplify
|
|
|
2d2418eb
|
2025-01-23T23:13:11
|
|
[VARC] Avoid float div-by-zero
|
|
|
fe9398e9
|
2025-01-23T22:59:12
|
|
[VARC] Chain transforms
Previous code was actually wrong even, for non-uniform x/y font scale.
Fixes https://github.com/harfbuzz/harfbuzz/issues/5017
|
|
|
de7f97b7
|
2024-11-15T15:14:37
|
|
[COLR] Add comment
|
|
6a25df24
|
2024-10-09T14:09:22
|
|
[COLR] Add and use get_clip_list ()
|
|
|
5462978c
|
2024-10-09T13:43:13
|
|
[COLR] Lets see if this makes CIFUZZ any happier
|
|
|
008505e1
|
2024-10-08T09:27:09
|
|
[COLR] Pepper some hb_barrier()'s around
The version is a minor version (because uint16), as such
version > 1 should be considered as supporting version 1
data.
|
|
|
cec95a2d
|
2024-10-07T21:06:56
|
|
Try to fix heap-buffer-overflow
|
|
|
4d1f6e04
|
2024-10-06T18:25:24
|
|
[COLR] Enable COLRv0 support in get_extents()
paint_glyph() supports COLRv0 but this code is never reached because of
the early return if version is not 1. This early return seems to be from
before COLRv0 extents were supported.
|
|
|
48774052
|
2024-09-22T13:00:09
|
|
[GDEF] Sprinkle some hb_barrier()s
|