f014a124f5
[set] Fix algebra
...
Was totally b0rked. Ouch!
Fixes https://github.com/harfbuzz/harfbuzz/issues/863
2018-03-07 10:49:26 +01:00
7587683cdc
[set] Minor
...
In preparation to fix https://github.com/harfbuzz/harfbuzz/issues/863
Binary operations are sets are completely broken. Ouch.
2018-03-07 09:57:11 +01:00
f18b9fbf65
[set] Implement iteration using bitop intrinsics
2018-02-16 18:14:41 -08:00
d25c3e69e9
[set] Readjust parameters
2018-02-16 17:45:09 -08:00
e1f1f96577
[set] Tune size again
...
Switch from a 128+64 hierarchy, to 32+32. New page size is 1024.
This should make set iteration faster, without affecting other op
performances much.
2018-02-14 15:44:48 -08:00
694eaf6367
[set] Add backwards iterator
...
New API:
- hb_set_previous()
- hb_set_previous_range()
2018-02-14 01:00:10 -08:00
fe3bc524bd
[set] Allow starting iteration from a non-member of the set
2018-02-13 23:51:45 -08:00
c479a59988
[subset] Assemble font
2018-02-07 21:26:04 -06:00
34ac3548b7
[set] Respect stride
...
Ouch!
2018-02-07 18:07:45 -06:00
c9e12a2b9b
Fix set initializer
2018-01-13 17:05:12 +00:00
2fe5f885b6
[set] Handle nil set in add_range() / add_sorted_array()
...
Fixes https://github.com/harfbuzz/harfbuzz/issues/657
2017-12-19 14:48:26 -05:00
493a005d95
[set] In add_sorted_array(), bail if data is not sorted
2017-12-16 11:49:39 -05:00
a7bd6d7a4c
[collect_glyphs] Bail if input data looks like garbage
...
Specificaly, when a range or sorted array has unexpected order, we take that as
font data being garbage and bail out. This fixes significant slowdown on a bad
version of Chandas font which has a 600KB GPOS with garbage inside.
Later on, I like to add a maximum-work counter for collect_glyphs to protect
against malicious fonts as well.
Fixes https://bugs.chromium.org/p/chromium/issues/detail?id=794896
2017-12-16 11:39:32 -05:00
1ce7d6e215
[set] Optimize add_array() / add_sorted_array()
...
Does page lookup as needed.
2017-12-16 11:39:11 -05:00
81f27df4d9
More work towards improving collect_glyphs() against bad input
...
The three "XXXXX"'s should be switched to false. Doing that separately for ease
of bisecting...
2017-12-16 06:12:06 -08:00
5d02572034
[set] Add add_sorted_array()
...
Not optimized to use sortedness yet. Also start putting in place infra
to faster reject bad data.
A version of Chandas.ttf found on some Chrome bots has 660kb of GPOS,
mostly junk. That is causing 48 million of set->add() calls in
collect_glyphs(), which is insane.
In the upcoming commits, I'll be speeding that up by optimizing
add_sorted_array(), while also reducing work by rejecting out-of-sort
arrays quickly and propagate the rejection.
Part of https://bugs.chromium.org/p/chromium/issues/detail?id=794896
2017-12-14 19:35:47 -08:00
9d6511a734
[set] Reduce number of preallocated pages from 8 to 1
...
Now that pagesize is 8192, this feels better.
2017-12-14 19:07:17 -08:00
9daa88cd79
Minor
2017-12-14 19:07:17 -08:00
f424a34223
[set] Change pagesize from 512 bits to 8192 bits
...
Fixes perf regression on some heavy fonts in Chrome's FT+HB
interaction.
See:
https://bugs.chromium.org/p/chromium/issues/detail?id=782220
More work to be done:
https://bugs.chromium.org/p/chromium/issues/detail?id=794896
2017-12-14 13:30:38 -08:00
0fe62c1f33
[set] Add add_array()
...
To be used to optimize adding a whole bunch of (sorted) items at the same time,
as in CoverageFormat1.
2017-12-14 10:39:00 -08:00
eeb26d21d4
[set] Actually commit fixes mentioned in previous commit!
2017-12-02 15:22:04 -08:00
0744149cda
[set] Protect against bad input in hb_set_add_range()
2017-12-02 15:06:15 -08:00
9d0194b3a8
[set] Optimize add_range() some more
...
It's as good as it gets, and seems to be on par with previous set implementation
in my benchmark.
Would be great if someone can double-check my bitops.
2017-12-01 13:56:06 -08:00
438c325a25
[set] Optimize add_range()
...
With new set implementation, this became really costy. Optimize it.
There's more to be done, but this shaves off most of the fat.
Part of fixing https://bugs.chromium.org/p/chromium/issues/detail?id=782220
2017-12-01 13:34:14 -08:00
dfd234a97d
[set] Fix page_map[] indexing in backward loop.
...
Fixes https://github.com/behdad/harfbuzz/issues/579 .
2017-10-26 12:33:05 -05:00
d511cfb557
[set] Bugfix for hb_set_next
2017-10-26 12:30:52 -05:00
30a591e3ce
[set] Rewrite another way :P
2017-10-23 14:29:20 -04:00
8170801d78
[set] Fix crash
...
Ouch! That's what happens when one plays with increment/decrement operators!
Fixes https://github.com/behdad/harfbuzz/issues/578
2017-10-23 14:29:14 -04:00
dd33e4e96b
[set] Don't use major()
...
Fixes https://github.com/behdad/harfbuzz/issues/577
2017-10-23 08:36:40 -04:00
d0f0ff85b9
[set] Don't use PAGE_SIZE
...
Fixes https://github.com/behdad/harfbuzz/issues/576
2017-10-23 08:34:30 -04:00
8b9d9b71b0
Fix set initialization issues
2017-10-22 17:48:06 -04:00
ba8b569607
Try fixing build on VC
...
c:\projects\harfbuzz\src\hb-set-private.hh(151): error C2327: 'hb_set_t::page_t::v': is not a type name, static, or enumerator [C:\projects\harfbuzz\build\harfbuzz.vcxproj]
2017-10-17 11:16:36 -07:00
a11249ee93
[set] Fix merge logic
2017-10-16 01:33:32 -04:00
ab8f327eb4
Ouch.
2017-10-15 18:21:35 -04:00
bb9917913f
[set] Disable vectorization
...
Fixes clang "non-const reference cannot bind to vector element" error.
2017-10-15 18:20:25 -04:00
0cf7d957b7
[set] Remove TODO items not worth pursuing
2017-10-15 17:01:31 -04:00
10d43658e7
[set] Speed up intersects()
2017-10-15 16:56:05 -04:00
7737e87ac4
Properly detect vector_size attribute and use fallback otherwise
2017-10-15 16:21:03 -04:00
f8a0ec50e2
[set] Add fallback implementation of int-vector type
2017-10-15 16:10:35 -04:00
deed4a48d1
Faster hb_set_t
...
Fixes https://github.com/behdad/harfbuzz/pull/23
2017-10-15 16:45:33 -03:00
a433e60a43
Remove unused hb_frozen_set_t
2017-10-15 14:09:46 +02:00
826a1daf2f
Move set-digests into their own header file
2017-10-15 14:09:05 +02:00
dbdbfe3d7b
Use nullptr instead of NULL
2017-10-15 12:11:08 +02:00
c3448e8d21
Use static_assert instead of custom ASSERT_STATIC
2017-10-15 12:02:00 +02:00
333173103b
Fix sign of shift operators
...
This one:
map->mask = (1 << (next_bit + bits_needed)) - (1 << next_bit);
before the fix, the shift was done as an int, causing overflow
if it ever got to 1 << 31. Sprinkle 'u's around.
Fixes https://bugs.chromium.org/p/chromium/issues/detail?id=634805
2016-08-08 17:28:14 -07:00
23e56e088a
Fix broken sentence
2015-07-25 17:38:21 +02:00
3fe4e92bc5
Minor
2015-04-08 12:49:23 -07:00
7b7129c7a9
Add hb_frozen_set_t
...
I experimented with replacing use of hb_set_digest_t with this new
hb_frozen_set_t, hoping to get a huge speedup for busy lookups
(like kern lookup in Roboto), but I only got 6% speendup in Roboto
and 4% in NotoNastaliqUrduDraft :(.
2015-02-25 15:43:25 -08:00
cd7ea4f791
Make hb_object_t members private
...
In preparation for fixing:
https://code.google.com/p/chromium/issues/detail?id=403594
2014-08-14 12:57:02 -04:00
2a8c49ade0
Remove unnecessary includes
2013-12-11 20:24:20 -05:00
Behdad Esfahbod
Jonathan Kew