blob: d8d219e436ecb8fec53cc6f9a30c5ef9b58a6bc8 [file] [log] [blame]
Overview of changes leading to 3.0.0
Friday, September 17, 2021
- Unicode 14.0 support (David Corbett).
- The hb-subset API and the harfbuzz-subset library's ABI are now declared
stable. The harfbuzz-subset library would not have been possible without the
work of Garret Rieger and Qunxin Liu from Google Fonts, and the earlier work
of Michiharu Ariza from Adobe.
- The hb-style API is now stable and no longer experimental.
- New API:
- Removed old unstable harfbuzz-subset API:
Overview of changes leading to 2.9.1
Tuesday, September 7, 2021
- Final subset API is in place and if no issues are discovered, it will be the
stable subset API of HarfBuzz 3.0.0. Old API is kept to ease transition, but
will be removed in 3.0.0.
- Various fuzzer-found bug fixes.
- hb_buffer_append() now handles the pre- and post-context which previously
were left unchanged in the destination buffer.
- hb-view / hb-shape now accept following new arguments:
o --unicodes-before/after: takes a list of hex numbers that represent Unicode
- Undeprecated API:
Overview of changes leading to 2.9.0
Wednesday, August 18, 2021
History Repeats Itself (Afghanistan)
- Subsetter API is being stabilized, with the first stable API to happen in
3.0.0 release (
- Support multiple variation axes with same tag, aka HOI.
- The “coretext” testing shaper now passes font variations to CoreText.
- hb-shape/hb-view does not break line at new lines unless text is read from
- hb-view and hb-subset has a --batch now, similar to hb-shape.
- The --batch mode now uses ; as argument separator instead of : used previously.
- The --batch in hb-shape does not expect 0th argument anymore. That is, the
lines read are interpreted as argv[1:], instead of argv[0:].
- The --batch option has been undocumented. We are ready to document it; send
feedback if you find it useful.
- hb-subset got arguments revamps. Added much-requested --gids-file, --glyphs,
--glyphs-file, --unicodes-file, supporting ranges in --unicodes.
- Various bug fixes.
Overview of changes leading to 2.8.2
Tuesday, July 8, 2021
- Shaping LTR digits for RTL scripts now makes the native direction of the
digits LTR, applying shaping and positioning rules on the same glyph order as
Uniscribe. (Jonathan Kew, Khaled Hosny).
- Subsetting COLR v1 and CPAL tables is now supported. (Garret Rieger, Qunxin Liu)
- Various fixes and improvements to the subsetter. (Garret Rieger, Qunxin Liu, Behdad)
- When applying morx table, mark glyph widths should not be zeroed. (Jonathan Kew)
- GPOS is preferred over kerx, if GSUB was applied. (Behdad)
- Regional_Indicator pairs are grouped together when clustering. (Behdad)
- New API:
Overview of changes leading to 2.8.1
Tuesday, May 4, 2021
- Subsetter now fully supports GSUB/GPOS/GDEF tables (including variations); as
such, layout tables are retained by subsetter by default. (Garret Rieger, Qunxin Liu)
- Build scripts no longer check for FontConfig as HarfBuzz does not use it.
- hb-view supports iTerm2 and kitty inline image protocols (Khaled Hosny),
it can also use Chafa for terminal graphics if available (Hans Petter Jansson).
Overview of changes leading to 2.8.0
Tuesday, March 16, 2021
- Shape joining scripts other than Arabic/Syriac using the Universal Shaping Engine.
Previously these were shaped using the generalized Arabic shaper. (David Corbett)
- Fix regression in shaping of U+0B55 ORIYA SIGN OVERLINE. (David Corbett)
- Update language tags. (David Corbett)
- Variations: reduce error: do not round each interpolated delta. (Just van Rossum)
- Documentation improvements. (Khaled Hosny, Nathan Willis)
- Subsetter improvements: subsets most, if not all, lookup types now. (Garret Rieger, Qunxin Liu)
- Fuzzer-found fixes and other improvements when memory failures happen. (Behdad)
- Removed most atomic implementations now that we have C++11 atomic impl. (Behdad)
- General codebase upkeep; using more C++11 features: constexpr constructors, etc. (Behdad)
Overview of changes leading to 2.7.4
Sunday, December 27, 2020
- Fix missing --enable-introspection configure option from previous release
- Documentation updates.
Overview of changes leading to 2.7.3
Wednesday, December 23, 2020
- Update USE shaper to 2020-08-13 specification, and other improvements.
- Don’t disable liga feature in myanmar shaper, to match Uniscribe.
- Improvements to language and script tags handling.
- Update language system tag registry to OpenType 1.8.4
- Support for serializing and deserializing Unicode buffers. Serialized buffers
are now delimited with `<>` or `[]` based on whether it is a Unicode or
glyphs buffer.
- Increase buffer work limits to handle fonts with many complex lookups.
- Handle more shaping operations in trace output.
- Memory access fixes.
- More OOM fixes.
- Improved documentation.
- Build system improvements.
- New API:
Overview of changes leading to 2.7.2
Saturday, August 29, 2020
- Fix a regression in the previous release that caused a crash with Kaithi.
- More OOM fixes.
Overview of changes leading to 2.7.1
Thursday, August 13, 2020
- ot-funcs now handles variable empty glyphs better when hvar/vvar isn't present.
- Reverted a GDEF processing regression.
- A couple of fixes to handle OOM better.
Overview of changes leading to 2.7.0
Saturday, July 25, 2020
- Use an implementation for round that always rounds up, some minor fluctuations
are expected on var font specially when hb-ot callback is used.
- Fix an AAT's `kerx` issue on broken rendering of Devanagari Sangam MN.
- Remove AAT's `lcar` table support from _get_ligature_carets API, not even much
use on macOS installed fonts (only two files). GDEF support is the recommended
one and expected to work properly after issues fixed two releases ago.
- Minor memory fixes to handle OOM better specially in hb-ft.
- Minor .so files versioning scheme change and remove stable/unstable scheme
differences, was never used in practice (always default to stable scheme).
- We are now suggesting careful packaging of the library using meson,
for more information.
- Distribution package URL is changed, either use GitHub generated tarballs,
or, even more preferably use commit hash of the release and git checkouts like,
Overview of changes leading to 2.6.8
Monday, June 22, 2020
- New API to fetch glyph alternates from GSUB table.
- hb-coretext build fix for macOS < 10.10.
- Meson build fixes, cmake port removal is postponed but please prepare for
it and give us feedback.
Autotools is still our main build system however please consider
experimenting with meson also for packaging the library.
- New API:
Overview of changes leading to 2.6.7
Wednesday, June 3, 2020
- Update to Unicode 13.0.0.
- Fix hb_ot_layout_get_ligature_carets for fonts without lcar table, it was
completely broken for all the other fonts since 2.1.2.
- As a part of our migration to meson, this release will be the last one
to provide cmake port files but autotools still is our main build system.
There is a possibility that the next version or the after be released
using meson.
Overview of changes leading to 2.6.6
Tuesday, May 12, 2020
- A fix in AAT kerning for Geeza Pro.
- Better support for resource fork fonts on macOS.
Overview of changes leading to 2.6.5
Friday, April 17, 2020
- Add experimental meson build system. Autotools is still the primary
and supported build system.
- AAT is now always preferred for horizontal scripts when both AAT and OT
layout tables exist at the same time.
- Subsetter improvements.
- New API:
Overview of changes leading to 2.6.4
Monday, October 29, 2019
- Small bug fix.
- Build fixes.
Overview of changes leading to 2.6.3
Monday, October 28, 2019
- Misc small fixes, mostly to build-related issues.
- New API:
Overview of changes leading to 2.6.2
Monday, September 30, 2019
- Misc small fixes, mostly to build-related issues.
Overview of changes leading to 2.6.1
Thursday, August 22, 2019
- Fix regression with hb_font_create_sub_font scaling introduced in 2.6.0.
- Change interpretation of font PTEM size / CoreText font size handling.
- hb-ot-font: Prefer symbol cmap subtable if present.
- Apply 'dist'/'abvm'/'blwm' features to all scripts.
- Drop experimental DirectWrite API.
Overview of changes leading to 2.6.0
Tuesday, August 13, 2019
- New OpenType metrics, baseline, and metadata table access APIs.
- New API to set font variations to a named-instance.
- New hb-gdi.h header and API for creating hb_face_t from HFONT.
- Amalgam: Provide a single-file file for easier alternate building.
- More size-reduction configurable options, enabled by HB_TINY.
- New API:
Overview of changes leading to 2.5.3
Wednesday, June 26, 2019
- Fix UCD script data for Unicode 10+ scripts. This was broken since 2.5.0.
- More optimizations for HB_TINY.
Overview of changes leading to 2.5.2
Thursday, June 20, 2019
- More hb-config.hh facilities to shrink library size, namely when built as
- New documentation of custom configurations in
- Fix build on gcc 4.8. That's supported again.
- Universal Shaping Engine improvements thanks to David Corbett.
- API Changes: Undeprecate some horizontal-kerning API and re-enable in hb-ft,
such that Type1 fonts will continue kerning.
Overview of changes leading to 2.5.1
Friday, May 31, 2019
- Fix build with various versions of Visual Studio.
- Improved documentation, thanks to Nathan Willis.
- Bugfix in subsetting glyf table.
- Improved scripts for cross-compiling for Windows using mingw.
A deprecated macro is added for backwards-compatibility.
Overview of changes leading to 2.5.0
Friday, May 24, 2019
- This release does not include much functional changes, but includes major internal
code-base changes. We now require C++11. Support for gcc 4.8 and earlier has been
- New hb-config.hh facility for compiling smaller library for embedded and web usecases.
- New Unicode Character Databse implementation that is half the size of previously-used
- Subsetter improvements.
- Improved documentation, thanks to Nathan Willis.
- Misc shaping fixes.
Overview of changes leading to 2.4.0
Monday, March 25, 2019
- Unicode 12.
- Misc fixes.
- Subsetter improvements.
- New API:
Overview of changes leading to 2.3.1
Wednesday, January 30, 2019
- AAT bug fixes.
- Misc internal housekeeping cleanup.
Overview of changes leading to 2.3.0
Thursday, December 20, 2018
- Fix regression on big-endian architectures. Ouch!
- Misc bug and build fixes.
- Fix subsetting of simple GSUB/GDEF.
- Merge CFF / CFF2 support contributed by Adobe. This mostly involves
the subsetter, but also get_glyph_extents on CFF fonts.
New API in hb-aat.h:
Overview of changes leading to 2.2.0
Thursday, November 29, 2018
- Misc shaping bug fixes.
- Add font variations named-instance API.
- Deprecate font variations axis enumeration API and add replacement.
- AAT shaping improvements:
o Fixed 'kern' table Format 2 implementation.
o Implement 'feat' table API for feature detection.
o Blacklist 'GSUB' table of fonts from 'MUTF' foundry that also have 'morx'.
New API:
Deprecated API:
Overview of changes leading to 2.1.3
Friday, November 16, 2018
- Fix AAT 'mort' shaping, which was broken in 2.1.2
Overview of changes leading to 2.1.2
Friday, November 16, 2018
- Various internal changes.
- AAT shaping improvements:
o Implement kern table Format 1 state-machine-based kerning.
o Implement cross-stream kerning (cursive positioning, etc).
o Ignore emptyish GSUB tables (zero scripts) if morx present.
o Don't apply GPOS if morx is being applied. Matches Apple.
-Overview of changes leading to 2.1.1
Monday, November 5, 2018
- AAT improvements:
o Implement 'mort' table.
o Implement 'kern' subtables Format 1 and Format 3.
Overview of changes leading to 2.1.0
Tuesday, October 30, 2018
- AAT shaping improvements:
o Allow user controlling AAT features, for whole buffer only currently.
o Several 'morx' fixes.
o Implement tuple-kerns in 'kerx'; Fixes kerning with Apple default
San Francisco fonts.
- Support for color fonts:
o COLR/CPAL API to fetch color layers.
o SVG table to fetch SVG documents.
o CBDT/sbix API to fetch PNG images.
- New 'name' table API.
- hb-ot-font now uses 'VORG' table to correctly position CFF glyphs
in vertical layout.
- Various fuzzer-found bug fixes.
Changed API:
A type and a macro added in 2.0.0 were renamed:
hb_name_id_t -> hb_ot_name_id_t
New API:
Overview of changes leading to 2.0.2
Saturday, October 20, 2018
- Fix two minor memory access issues in AAT tables.
Overview of changes leading to 2.0.1
Friday, October 19, 2018
- Fix hb-version.h reported release version that went wrong (1.8.0)
with previous release.
- Fix extrapolation in 'trak' table.
- Fix hb-font infinite-recursion issue with some font funcs and
subclassed fonts.
- Implement variation-kerning format in kerx table, although without
- Fix return value of hb_map_is_empty().
Overview of changes leading to 2.0.0
Thursday, October 18, 2018
- Added AAT shaping support (morx/kerx/trak).
Automatically used if GSUB/GPOS are not available respectively.
Set HB_OPTIONS=aat env var to have morx/kerx preferred over
- Apply TrueType kern table internally, instead of relying on
hb_font_t callbacks.
- Khmer shaper significantly rewritten to better match Uniscribe.
- Indic3 tags ('dev3', etc) are passed to USE shaper.
- .dfont Mac font containers implemented.
- Script- and language-mapping revamped to better use BCP 47.
- Misc USE and Indic fixes.
- Misc everything fixes.
- Too many things to list. Biggest release since 0.9.1, with
over 500 commits in just over 5 weeks! Didn't intend it to
be a big release. Just happened to become.
- hb-ft now locks underlying FT_Face during use.
API changes:
- Newly-created hb_font_t's now have our internal "hb-ot-font"
callbacks set on them, so they should work out of the box
without any callbacks set. If callbacks are set, everything
is back to what it was before, the fallback callbacks are
null. If you to get the internal implementation modified,
sub_font it.
- New hb_font_funcs_set_nominal_glyphs_func() allows speeding
up character to glyph mapping.
New API:
Deprecated API:
Overview of changes leading to 1.9.0
Monday, September 10, 2018
- Added 'cmap' API to hb_face_t.
- Face-builder API.
- hb-ot-font re-creation should be much leaner now, as the
font tables it uses are cached on hb_face_t now.
- Internal source header file name changes:
hb-*-private.hh is renamed to hb-*.hh.
New API:
Overview of changes leading to 1.8.8
Tuesday, August 14, 2018
- Fix hb-icu crash on architectures where compare_exchange_weak() can
fail falsely. This bug was introduced in 1.8.4.
- More internal refactoring of atomic operations and singletons.
- API changes:
The following functions do NOT reference their return value before
* hb_unicode_funcs_get_default()
* hb_glib_get_unicode_funcs()
* hb_icu_get_unicode_funcs()
This is consistent with their naming ("get", instead of "reference")
as well as how they are used in the wild (ie. no one calls destroy()
on their return value.)
Overview of changes leading to 1.8.7
Wednesday, August 8, 2018
- Fix assertion failure with GDEF-blacklisted fonts.
Overview of changes leading to 1.8.6
Tuesday, August 7, 2018
- Internal code shuffling.
- New API to speed up getting advance widths for implementations
that have heavy overhead in get_h_advance callback:
Overview of changes leading to 1.8.5
Wednesday, August 1, 2018
- Major Khmer shaper improvements to better match Microsoft.
- Indic bug fixes.
- Internal improvements to atomic operations.
Overview of changes leading to 1.8.4
Tuesday, July 17, 2018
- Fix build on non-C++11.
- Use C++-style GCC atomics and C++11 atomics.
Overview of changes leading to 1.8.3
Wednesday, July 11, 2018
- A couple of Indic / USE bug fixes.
- Disable vectorization, as it was causing unaligned access bus error on
certain 32bit architectures.
Overview of changes leading to 1.8.2
Tuesday, July 3, 2018
- Fix infinite loop in Khmer shaper.
- Improve hb_blob_create_from_file() for streams.
Overview of changes leading to 1.8.1
Tuesday, June 12, 2018
- Fix hb-version.h file generation; last two releases went out with wrong ones.
- Add correctness bug in hb_set_t operations, introduced in 1.7.7.
- Remove HB_SUBSET_BUILTIN build option. Not necessary.
Overview of changes leading to 1.8.0
Tuesday, June 5, 2018
- Update to Unicode 11.0.0.
Overview of changes leading to 1.7.7
Tuesday, June 5, 2018
- Lots of internal changes, but not yet exposed externally.
- All HarfBuzz objects are significantly smaller in size now.
- Sinhala: Position repha on top of post-consonant, not base.
This better matches Windows 10 behavior, which was changed
from previous Windows versions.
- New build options:
o New cpp macro HB_NO_ATEXIT
o New cpp macro HB_SUBSET_BUILTIN
- Significant libharfbuzz-subset changes. API subject to change.
- New API in libharfbuzz:
A hashmap implementation:
Overview of changes leading to 1.7.6
Wednesday, March 7, 2018
- Fix to hb_set_t binary operations. Ouch.
- New experimental harfbuzz-subset library. All of hb-subset.h
is experimental right now and API WILL change.
- New API:
Overview of changes leading to 1.7.5
Tuesday, January 30, 2018
- Separate Khmer shaper from Indic.
- First stab at AAT morx. Not hooked up.
- Misc bug fixes.
Overview of changes leading to 1.7.4
Wednesday, December 20, 2017
- Fix collect_glyphs() regression caused by hb_set_t changes.
Overview of changes leading to 1.7.3
Monday, December 18, 2017
- hb_set_t performance tuning and optimizations.
- Speed up collect_glyphs() and reject garbage data.
- In hb_coretext_font_create() set font point-size (ptem).
- Misc fixes.
Overview of changes leading to 1.7.2
Monday, December 4, 2017
- Optimize hb_set_add_range().
- Misc fixes.
- New API:
Overview of changes leading to 1.7.1
Tuesday, November 14, 2017
- Fix atexit object destruction regression.
- Fix minor integer-overflow.
Overview of changes leading to 1.7.0
Monday, November 13, 2017
- Minor Indic fixes.
- Implement kerning and glyph names in hb-ot-font.
- Various DSO optimization re .data and .bss sizes.
- Make C++11 optional; build fixes.
- Mark all other backends "unsafe-to-break".
- Graphite fix.
Overview of changes leading to 1.6.3
Thursday, October 26th, 2017
- Fix hb_set_t some more. Should be solid now.
- Implement get_glyph_name() for hb-ot-font.
- Misc fixes.
Overview of changes leading to 1.6.2
Monday, October 23nd, 2017
- Yesterday's release had a bad crasher; don't use it. That's what
happens when one works on Sunday...
- Build fixes for FreeBSD and Chrome Android.
Overview of changes leading to 1.6.1
Sunday, October 22nd, 2017
- Don't skip over COMBINING GRAPHEME JOINER when ligating, etc.
To be refined:
- Faster hb_set_t implementation.
- Don't use deprecated ICU API.
- Fix undefined-behavior in Myanmar shaper, introduced in 1.6.0
- Deprecated API:
Overview of changes leading to 1.6.0
Friday, October the 13th, 2017
- Update to Unicode 10.
- Various Indic and Universal Shaping Engine fixes as a result of
HarfBuzz Hackfest with Jonathan Kew at Web Engines Hackfest at
the Igalia offices in A Coruña, Spain. Thanks Igalia for having
- Implement Unicode Arabic Mark Ordering Algorithm UTR#53.
- Implement optical sizing / tracking in CoreText backend, using
new API hb_font_set_ptem().
- Allow notifying hb_font_t that underlying FT_Face changed sizing,
using new API hb_ft_font_changed().
- More Graphite backend RTL fixes.
- Fix caching of variable font shaping plans.
- hb-view / hb-shape now accept following new arguments:
o --unicodes: takes a list of hex numbers that represent Unicode
New API:
Overview of changes leading to 1.5.1
Tuesday, September 5, 2017
- Fix "unsafe-to-break" in fallback shaping and other corner cases.
All our tests pass with --verify now, meaning unsafe-to-break API
works as expected.
- Add --unicodes to hb-view / hb-shape.
- [indic] Treat Consonant_With_Stacker as consonant. This will need
further tweaking.
- hb_buffer_diff() tweaks.
Overview of changes leading to 1.5.0
Wednesday, August 23, 2017
- Misc new API, for appending a buffer to another, and for comparing
contents of two buffers for types of differences.
- New "unsafe-to-break" API. Can be used to speed up reshaping
in line-breaking situations. Essentially, after shaping, it returns
positions in the input string (some of the cluster boundaries) that
are "safe to break" in that if the text is segmented at that position
and two sides reshaped and concatenated, the shaping result is
exactly the same as shaping the text in one piece.
hb-view and hb-shape and hb-shape now take --verify, which verifies
the above property.
Some corner cases of the implementation are still not quite working.
Those will be fixed in subsequent releases.
- New API:
Overview of changes leading to 1.4.8
Tuesday, August 8, 2017
- Major fix to avar table handling.
- Rename hb-shape --show-message to --trace.
- Build fixes.
Overview of changes leading to 1.4.7
Tuesday, July 18, 2017
- Multiple Indic, Tibetan, and Cham fixes.
- CoreText: Allow disabling kerning.
- Adjust Arabic feature order again.
- Misc build fixes.
Overview of changes leading to 1.4.6
Sunday, April 23, 2017
- Graphite2: Fix RTL positioning issue.
- Backlist GDEF of more versions of Padauk and Tahoma.
- New, experimental, cmake alternative build system.
Overview of changes leading to 1.4.5
Friday, March 10, 2017
- Revert "Fix Context lookup application when moving back after a glyph..."
This introduced memory access problems. To be fixed properly soon.
Overview of changes leading to 1.4.4
Sunday, March 5, 2017
- Fix Context lookup application when moving back after a glyph deletion.
- Fix buffer-overrun in Bengali.
Overview of changes leading to 1.4.3
Saturday, February 25, 2017
- Route Adlam script to Arabic shaper.
- Misc fixes.
- New API:
- Deprecate API:
Overview of changes leading to 1.4.2
Monday, January 23, 2017
- Implement OpenType Font Variation tables avar/fvar/HVAR/VVAR.
- hb-shape and hb-view now accept --variations.
- New API:
- MVAR to be implemented later. Access to named instances to be
implemented later as well.
- Misc fixes.
Overview of changes leading to 1.4.1
Thursday, January 5, 2017
- Always build and use UCDN for Unicode data by default.
Reduces dependence on version of Unicode data in glib,
specially in the Windows bundles we are shipping, which
have very old glib.
Overview of changes leading to 1.4.0
Thursday, January 5, 2017
- Merged "OpenType GX" branch which adds core of support for
OpenType 1.8 Font Variations. To that extent, the relevant
new API is:
New API:
with supporting API:
New API:
Currently variations in GSUB/GPOS/GDEF are fully supported,
and no other tables are supported. In particular, fvar/avar
are NOT supported, hence the hb_font_set_var_coords_normalized()
taking normalized coordinates. API to take design coordinates
will be added in the future.
HVAR/VVAR/MVAR support will also be added to hb-ot-font in the
- Fix regression in GDEF glyph class processing.
- Add decompositions for Chakma, Limbu, and Balinese in USE shaper.
- Misc fixes.
Overview of changes leading to 1.3.4
Monday, December 5, 2016
- Fix vertical glyph origin in hb-ot-font.
- Implement CBDT/CBLC color font glyph extents in hb-ot-font.
Overview of changes leading to 1.3.3
Wednesday, September 28, 2016
- Implement parsing of OpenType MATH table.
New API:
Overview of changes leading to 1.3.2
Wednesday, September 27, 2016
- Fix build of hb-coretext on older OS X versions.
Overview of changes leading to 1.3.1
Wednesday, September 7, 2016
- Blacklist bad GDEF of more fonts (Padauk).
- More CoreText backend crash fixes with OS X 10.9.5.
- Misc fixes.
Overview of changes leading to 1.3.0
Thursday, July 21, 2016
- Update to Unicode 9.0.0
- Move Javanese from Indic shaper to Universal Shaping Engine.
- Allow MultipleSubst to delete a glyph (matching Windows engine).
- Update Universal Shaping Engine to latest draft from Microsoft.
- DirectWrite backend improvements. Note: this backend is for testing ONLY.
- CoreText backend improvements with unreachable fonts.
- Implement symbol fonts (cmap 3.0.0) in hb-ft and hb-ot-font.
- Blacklist bad GDEF of more fonts (Tahoma & others).
- Misc fixes.
Overview of changes leading to 1.2.7
Monday, May 2, 2016
- Blacklist another version of Times New Roman (Bold) Italic from Windows 7.
- Fix Mongolian Free Variation Selectors shaping with certain fonts.
- Fix Tibetan shorthand contractions shaping.
- Improved list of language tag mappings.
- Unbreak build on Windows CE.
- Make 'glyf' table loading lazy in hb-ot-font.
Overview of changes leading to 1.2.6
Friday, April 8, 2016
- Blacklist GDEF table of another set of Times New Roman (Bold) Italic.
- DirectWrite backend improvements. Note: DirectWrite backend is
exclusively for our internal testing and should NOT be used in any
production system whatsoever.
Overview of changes leading to 1.2.5
Monday, April 4, 2016
- Fix GDEF mark-filtering-set, which was broken in 1.2.3.
Overview of changes leading to 1.2.4
Thursday, March 17, 2016
- Synthesize GDEF glyph class for any glyph that does not have one in GDEF.
I really hope we don't discover broken fonts that shape badly with this
- Misc build and other minor fixes.
- API changes:
- Added HB_NDEBUG. It's fine for production systems to define this to
disable high-overhead debugging checks. However, I also reduced the
overhead of those checks, so it's a non-issue right now. You can
forget it. Just not defining anything at all is fine.
Overview of changes leading to 1.2.3
Thursday, February 25, 2016
- Blacklist GDEF table of certain versions of Times New Roman (Bold) Italic,
due to bug in glyph class of ASCII double-quote character. This should
address "regression" introduced in 1.2.0 when we switched mark zeroing
in most shapers from BY_UNICODE_LATE to BY_GDEF_LATE.
This fourth release in a week should finally stablize things...
- hb-ot-font's get_glyph() implementation saw some optimizations. Though,
might be really hard to measure in real-world situations.
- Also, two rather small API changes:
We now disable some time-consuming internal bookkeeping if built with NDEBUG
defined. This is a first time that we use NDEBUG to disable debug code. If
there exist production systems that do NOT want to enable NDEBUG, please let
me know and I'll add HB_NDEBUG.
Added get_nominal_glyph() and get_variation_glyph() instead of get_glyph()
New API:
- hb_font_get_nominal_glyph_func_t
- hb_font_get_variation_glyph_func_t
- hb_font_funcs_set_nominal_glyph_func()
- hb_font_funcs_set_variation_glyph_func()
- hb_font_get_nominal_glyph()
- hb_font_get_variation_glyph()
Deprecated API:
- hb_font_get_glyph_func_t
- hb_font_funcs_set_glyph_func()
Clients that implement their own font-funcs are encouraged to replace
their get_glyph() implementation with a get_nominal_glyph() and
get_variation_glyph() pair. The variation version can assume that
variation_selector argument is not zero. Old (deprecated) functions
will continue working indefinitely using internal gymnastics; it is
just more efficient to use the new functions.
Overview of changes leading to 1.2.2
Wednesday, February 24, 2016
- Fix regression with mark positioning with fonts that have
non-zero mark advances. This was introduced in 1.2.0 while
trying to make mark and cursive attachments to work together.
I have partially reverted that, so this version is much more
like what we had before. All clients who updated to 1.2.0
should update to this version.
Overview of changes leading to 1.2.1
Tuesday, February 23, 2016
- CoreText: Fix bug with wrong scale if font scale was changed later.
- CoreText: Drastically speed up font initialization.
- CoreText: Fix tiny leak.
- Group ZWJ/ZWNJ with previous syllable under cluster-level=0.
- Add test/shaping/ about how to add tests to the suite.
Overview of changes leading to 1.2.0
Friday, February 19, 2016
- Fix various issues (hangs mostly) in case of memory allocation failure.
- Change mark zeroing types of most shapers from BY_UNICODE_LATE to
BY_GDEF_LATE. This seems to be what Uniscribe does.
- Change mark zeroing of USE shaper from NONE to BY_GDEF_EARLY. That's
what Windows does.
- Allow GPOS cursive connection on marks, and fix the interaction with
mark attachment. This work resulted in some changes to how mark
attachments work. See:
- Graphite2 shaper: improved negative advance handling (eg. Nastaliq).
- Add nmake-based build system for Windows.
- Minor speedup.
- Misc. improvements.
Overview of changes leading to 1.1.3
Monday, January 11, 2016
- Ported Indic shaper to Unicode 8.0 data.
- Universal Shaping Engine fixes.
- Speed up CoreText shaper when font fallback happens in CoreText.
- Documentation improvements, thanks to Khaled Hosny.
- Very rough directwrite shaper for testing, thanks to Ebrahim Byagowi.
- Misc bug fixes.
- New API:
* Font extents:
* Buffer message (aka debug):
Actual message protocol to be fleshed out later.
Overview of changes leading to 1.1.2
Wednesday, November 26, 2015
- Fix badly-broken fallback shaper that affected terminology.
- Fix y_scaling in Graphite shaper.
- API changes:
* An unset glyph_h_origin() function in font-funcs now (sensibly)
implies horizontal origin at 0,0. Ie, the nil callback returns
true instead of false. As such, implementations that have a
glyph_h_origin() that simply returns true, can remove that function
with HarfBuzz >= 1.1.2. This results in a tiny speedup.
Overview of changes leading to 1.1.1
Wednesday, November 24, 2015
- Build fixes, specially for hb-coretext.
Overview of changes leading to 1.1.0
Wednesday, November 18, 2015
- Implement 'stch' stretch feature for Syriac Abbreviation Mark.
- Disable use of decompose_compatibility() callback.
- Implement "shaping" of various Unicode space characters, even
if the font does not support them.
- If font does not support U+2011 NO-BREAK HYPHEN, fallback to
U+2010 HYPHEN.
- Changes resulting from libFuzzer continuous fuzzing:
* Reject font tables that need more than 8 edits,
* Bound buffer growth during shaping to 32x,
* Fix assertions and other issues at OOM / buffer max-growth.
- Misc fixes and optimizations.
- API changes:
* All fonts created with hb_font_create() now inherit from
(ie. have parent) hb_font_get_empty().
Overview of changes leading to 1.0.6
Thursday, October 15, 2015
- Reduce max nesting level in OT lookups from 8 to 6.
Should not affect any real font as far as I know.
- Fix memory access issue in ot-font.
- Revert default load-flags of fonts created using hb_ft_font_create()
back to FT_LOAD_DEFAULT|FT_LOAD_NO_HINTING. This was changed in
last release (1.0.5), but caused major issues, so revert.
Overview of changes leading to 1.0.5
Tuesday, October 13, 2015
- Fix multiple memory access bugs discovered using libFuzzer.
Everyone should upgrade to this version as soon as possible.
We now have continuous fuzzing set up, to avoid issues like
these creeping in again.
- Misc fixes.
- New API:
* hb_font_set_parent().
* hb_ft_font_[sg]et_load_flags()
The default flags for fonts created using hb_ft_font_create()
has changed to default to FT_LOAD_DEFAULT now. Previously it
- API changes:
* Fonts now default to units-per-EM as their scale, instead of 0.
* hb_font_create_sub_font() does NOT make parent font immutable
anymore. hb_font_make_immutable() does.
Overview of changes leading to 1.0.4
Wednesday, September 30, 2015
- Fix minor out-of-bounds read error.
Overview of changes leading to 1.0.3
Tuesday, September 1, 2015
- Start of user documentation, from Simon Cozens!
- Implement glyph_extents() for TrueType fonts in hb-ot-font.
- Improve GPOS cursive attachments with conflicting lookups.
- More fixes for cluster-level = 1.
- Uniscribe positioning fix.
Overview of changes leading to 1.0.2
Wednesday, August 19, 2015
- Fix shaping with cluster-level > 0.
- Fix Uniscribe backend font-size scaling.
- Declare dependencies in harfbuzz.pc.
FreeType is not declared though, to avoid bugs in pkg-config
0.26 with recursive dependencies.
- Slightly improved debug infrastructure. More to come later.
- Misc build fixes.
Overview of changes leading to 1.0.1
Monday, July 27, 2015
- Fix out-of-bounds access in USE shaper.
Overview of changes leading to 1.0.0
Sunday, July 26, 2015
- Implement Universal Shaping Engine:
- Bump version to 1.0.0. The soname was NOT bumped.
Overview of changes leading to 0.9.42
Thursday, July 26, 2015
- New API to allow for retrieving finer-grained cluster
mappings if the client desires to handle them. Default
behavior is unchanged.
- Fix cluster merging when removing default-ignorables.
- Update to Unicode 8.0
- hb-graphite2 fixes.
- Misc fixes.
- Removed HB_NO_MERGE_CLUSTERS hack.
- New API:
hb_buffer_cluster_level_t enum
hb-shape / hb-view --cluster-level
Overview of changes leading to 0.9.41
Thursday, June 18, 2015
- Fix hb-coretext with trailing whitespace in right-to-left.
- New API: hb_buffer_reverse_range().
- Allow implementing atomic ops in config.h.
- Fix hb_language_t in language bindings.
- Misc fixes.
Overview of changes leading to 0.9.40
Friday, March 20, 2015
- Another hb-coretext crasher fix. Ouch!
- Happy Norouz!
Overview of changes leading to 0.9.39
Wednesday, March 4, 2015
- Critical hb-coretext fixes.
- Optimizations and refactoring; no functional change
- Misc build fixes.
Overview of changes leading to 0.9.38
Friday, January 23, 2015
- Fix minor out-of-bounds access in Indic shaper.
- Change New Tai Lue shaping engine from South-East Asian to default,
reflecting change in Unicode encoding model.
- Add hb-shape --font-size. Can take up to two numbers for separate
x / y size.
- Fix CoreText and FreeType scale issues with negative scales.
- Reject blobs larger than 2GB. This might break some icu-le-hb clients
that need security fixes. See:
- Avoid accessing font tables during face destruction, in casce rogue
clients released face data already.
- Fix up gobject-introspection a bit. Python bindings kinda working.
See README.python.
- Misc fixes.
- API additions:
Overview of changes leading to 0.9.37
Wednesday, December 17, 2014
- Fix out-of-bounds access in Context lookup format 3.
- Indic: Allow ZWJ/ZWNJ before syllable modifiers.
Overview of changes leading to 0.9.36
Thursday, November 20, 2014
- First time that three months went by without a release since
0.9.2 was released on August 10, 2012!
- Fix performance bug in hb_ot_collect_glyphs():
- Add basic vertical-text support to hb-ot-font.
- Misc build fixes.
Overview of changes leading to 0.9.35
Saturday, August 13, 2014
- Fix major shape-plan caching bug when more than one shaper were
provided to hb_shape_full() (as exercised by XeTeX).
- Fix Arabic fallback shaping regression. This was broken in 0.9.32.
- Major hb-coretext fixes. That backend is complete now, including
respecing buffer direction and language, down to vertical writing.
- Build fixes for Windows CE. Should build fine now.
- Misc fixes:
Use atexit() only if it's safe to call from shared library
Mandaic had errors in its Unicode Joining_Type
- API changes:
* hb_buffer_clear_contents() does not reset buffer flags now.
After 763e5466c0a03a7c27020e1e2598e488612529a7, one doesn't
need to set flags for different pieces of text. The flags now
are something the client sets up once, depending on how it
actually uses the buffer. As such, don't clear it in
I don't expect any changes to be needed to any existing client.
Overview of changes leading to 0.9.34
Saturday, August 2, 2014
- hb_feature_from_string() now accepts CSS font-feature-settings format.
- As a result, hb-shape / hb-view --features also accept CSS-style strings.
Eg, "'liga' off" is accepted now.
- Add old-spec Myanmar shaper:
- Don't apply 'calt' in Hangul shaper.
- Fix mark advance zeroing for Hebrew shaper:
- Implement Windows-1256 custom Arabic shaping. Only built on Windows,
and requires help from get_glyph(). Used by Firefox.
- Disable 'liga' in vertical text.
- Build fixes.
- API changes:
* Make HB_BUFFER_FLAG_BOT/EOT easier to use.
Previously, we expected users to provide BOT/EOT flags when the
text *segment* was at paragraph boundaries. This meant that for
clients that provide full paragraph to HarfBuzz (eg. Pango), they
had code like this:
hb_buffer_set_flags (hb_buffer,
(item_offset == 0 ? HB_BUFFER_FLAG_BOT : 0) |
(item_offset + item_length == paragraph_length ?
hb_buffer_add_utf8 (hb_buffer,
paragraph_text, paragraph_length,
item_offset, item_length);
After this change such clients can simply say:
hb_buffer_set_flags (hb_buffer,
hb_buffer_add_utf8 (hb_buffer,
paragraph_text, paragraph_length,
item_offset, item_length);
Ie, HarfBuzz itself checks whether the segment is at the beginning/end
of the paragraph. Clients that only pass item-at-a-time to HarfBuzz
continue not setting any flags whatsoever.
Another way to put it is: if there's pre-context text in the buffer,
HarfBuzz ignores the BOT flag. If there's post-context, it ignores
EOT flag.
Overview of changes leading to 0.9.33
Tuesday, July 22, 2014
- Turn off ARabic 'cswh' feature that was accidentally turned on.
- Make hb_face_make_immutable() really make face immutable!
- Windows build fixes.
Overview of changes leading to 0.9.32
Thursday, July 17, 2014
- Apply Arabic shaping features in spec order exactly.
- Another fix for Mongolian free variation selectors.
- For non-Arabic scripts in Arabic shaper apply 'rlig' and 'calt'
- Minor adjustment to U+FFFD logic.
- Fix hb-coretext build.
Overview of changes leading to 0.9.31
Wednesday, July 16, 2014
- Only accept valid UTF-8/16/32; we missed many cases before.
- Better shaping of invalid UTF-8/16/32. Falls back to
- With all changes in this release, the buffer will contain fully
valid Unicode after hb_buffer_add_utf8/16/32 no matter how
broken the input is. This can be overridden though. See below.
- Fix Mongolian Variation Selectors for fonts without GDEF.
- Fix minor invalid buffer access.
- Accept zh-Hant and zh-Hans language tags. hb_ot_tag_to_language()
now uses these instead of private tags.
- Build fixes.
- New API:
* hb_buffer_add_codepoints(). This does what hb_buffer_add_utf32()
used to do, ie. no validity check on the input at all. add_utf32
now replaces invalid Unicode codepoints with the replacement
character (see below).
* hb_buffer_set_replacement_codepoint()
* hb_buffer_get_replacement_codepoint()
Previously, in hb_buffer_add_utf8 and hb_buffer_add_utf16, when
we detected broken input, we replaced that with (hb_codepoint_t)-1.
This has changed to use U+FFFD now, but can be changed using these
new API.
Overview of changes leading to 0.9.30
Wednesday, July 9, 2014
- Update to Unicode 7.0.0:
* New scripts Manichaean and Psalter Pahlavi are shaped using
Arabic shaper.
* All the other new scripts to through the generic shaper for
- Minor Indic improvements.
- Fix graphite2 backend cluster mapping [crasher!]
- API changes:
* New HB_SCRIPT_* values for Unicode 7.0 scripts.
* New function hb_ot_layout_language_get_required_feature().
- Build fixes.
Overview of changes leading to 0.9.29
Thursday, May 29, 2014
- Implement cmap in hb-ot-font.h. No variation-selectors yet.
- Myanmar: Allow MedialYa+Asat.
- Various Indic fixes:
* Support most characters in Extended Devanagary and Vedic
Unicode blocks.
* Allow digits and a some punctuation as consonant placeholders.
- Build fixes.
Overview of changes leading to 0.9.28
Monday, April 28, 2014
- Unbreak old-spec Indic shaping. (bug 76705)
- Fix shaping of U+17DD and U+0FC6.
- Add HB_NO_MERGE_CLUSTERS build option. NOT to be enabled by default
for shipping libraries. It's an option for further experimentation
right now. When we are sure how to do it properly, we will add
public run-time API for the functionality.
- Build fixes.
Overview of changes leading to 0.9.27
Tuesday, March 18, 2014
- Don't use "register" storage class specifier
- Wrap definition of free_langs() with HAVE_ATEXIT
- Add coretext_aat shaper and hb_coretext_face_create() constructor
- If HAVE_ICU_BUILTIN is defined, use hb-icu Unicode callbacks
- Add Myanmar test case from OpenType Myanmar spec
- Only do fallback Hebrew composition if no GPOS 'mark' available
- Allow bootstrapping without gtk-doc
- Use AM_MISSING_PROG for ragel and git
- Typo in ucdn's
- Improve MemoryBarrier() implementation
Overview of changes leading to 0.9.26
Thursday, January 30, 2014
- Misc fixes.
- Fix application of 'rtlm' feature.
- Automatically apply frac/numr/dnom around U+2044 FRACTION SLASH.
- New header: hb-ot-shape.h
- Uniscribe: fix scratch-buffer accounting.
- Reorder Tai Tham SAKOT to after tone-marks.
- Add Hangul shaper.
- New files:
- Disable 'cswh' feature in Arabic shaper.
- Coretext: better handle surrogate pairs.
Overview of changes leading to 0.9.25
Wednesday, December 4, 2013
- Myanmar shaper improvements.
- Avoid font fallback in CoreText backend.
- Additional OpenType language tag mappiongs.
- More aggressive shape-plan caching.
- Build with / require automake 1.13.
- Build with libtool alpha to support ppc64le.
Overview of changes leading to 0.9.24
Tuesday, November 13, 2013
- Misc compiler warning fixes with clang.
- No functional changes.
Overview of changes leading to 0.9.23
Monday, October 28, 2013
- "Udupi HarfBuzz Hackfest", Paris, October 14..18 2013.
- Fix (Chain)Context recursion with non-monotone lookup positions.
- Misc Indic bug fixes.
- New Javanese / Buginese shaping, similar to Windows 8.1.
Overview of changes leading to 0.9.22
Thursday, October 3, 2013
- Fix use-after-end-of-scope in hb_language_from_string().
- Fix hiding of default_ignorables if font doesn't have space glyph.
- Protect against out-of-range lookup indices.
- API Changes:
* Added hb_ot_layout_table_get_lookup_count()
Overview of changes leading to 0.9.21
Monday, September 16, 2013
- Rename gobject-introspection library name from harfbuzz to HarfBuzz.
- Remove (long disabled) hb-old and hb-icu-le test shapers.
- Misc gtk-doc and gobject-introspection annotations.
- Misc fixes.
- API changes:
Overview of changes leading to 0.9.20
Thursday, August 29, 2013
- Misc substitute_closure() fixes.
- Build fixes.
- gtk-doc boilerplate integrated. Docs are built now, but
contain no contents. By next release hopefully we have
some content in. Enable using --enable-gtk-doc.
GObject and Introspection:
- Added harfbuzz-gobject library (hb-gobject.h) that has type
bindings for all HarfBuzz objects and enums. Enable using
- Added gobject-introspection boilerplate. Nothing useful
right now. Work in progress. Gets enabled automatically if
--with-gobject is used. Override with --disable-introspection.
OpenType shaper:
- Apply 'mark' in Myanmar shaper.
- Don't apply 'dlig' by default.
Uniscribe shaper:
- Support user features.
- Fix loading of fonts that are also installed on the system.
- Fix shaping of Arabic Presentation Forms.
- Fix build with wide chars.
CoreText shaper:
- Support user features.
Source changes:
- hb_face_t code moved to hb-face.h /
- Added hb-deprecated.h.
API changes:
- Deprecated HB_SCRIPT_CANADIAN_ABORIGINAL; replaced by
- Deprecated HB_BUFFER_FLAGS_DEFAULT; replaced by
Overview of changes leading to 0.9.19
Tuesday, July 16, 2013
- Build fixes.
- Better handling of multiple variation selectors in a row.
- Pass on variation selector to GSUB if not consumed by cmap.
- Fix undefined memory access.
- Add Javanese config to Indic shaper.
- Misc bug fixes.
Overview of changes leading to 0.9.18
Tuesday, May 28, 2013
New build system:
- All unneeded code is all disabled by default,
- Uniscribe and CoreText shapers can be enabled with their --with options,
- icu_le and old shapers cannot be enabled for now,
- glib, freetype, and cairo will be detected automatically.
They can be force on/off'ed with their --with options,
- icu and graphite2 are default off, can be enabled with their --with
Moreover, ICU support is now build into a separate library:, and a new harfbuzz-icu.pc is shipped for it.
Distros can enable ICU now without every application on earth
getting linked to via
For distros I recommend that they make sure they are building --with-glib
--with-freetype --with-cairo, --with-icu, and optionally --with-graphite2;
And package harfbuzz and harfbuzz-icu separately.
Overview of changes leading to 0.9.17
Monday, May 20, 2013
- Build fixes.
- Fix bug in hb_set_get_min().
- Fix regression with Arabic mark positioning / width-zeroing.
Overview of changes leading to 0.9.16
Friday, April 19, 2013
- Major speedup in OpenType lookup processing. With the Amiri
Arabic font, this release is over 3x faster than previous
release. All scripts / languages should see this speedup.
- New --num-iterations option for hb-shape / hb-view; useful for
Overview of changes leading to 0.9.15
Friday, April 05, 2013
- Build fixes.
- Fix crasher in graphite2 shaper.
- Fix Arabic mark width zeroing regression.
- Don't compose Hangul jamo into Unicode syllables.
Overview of changes leading to 0.9.14
Thursday, March 21, 2013
- Build fixes.
- Fix time-consuming sanitize with malicious fonts.
- Implement hb_buffer_deserialize_glyphs() for both json and text.
- Do not ignore Hangul filler characters.
- Indic fixes:
* Fix Malayalam pre-base reordering interaction with post-forms.
* Further adjust ZWJ handling. Should fix known regressions from
Overview of changes leading to 0.9.13
Thursday, February 25, 2013
- Build fixes.
- Ngapi HarfBuzz Hackfest in London (February 2013):
* Fixed all known Indic bugs,
* New Win8-style Myanmar shaper,
* New South-East Asian shaper for Tai Tham, Cham, and New Tai Lue,
* Smartly ignore Default_Ignorable characters (joiners, etc) wheb
matching GSUB/GPOS lookups,
* Fix 'Phags-Pa U+A872 shaping,
* Fix partial disabling of default-on features,
* Allow disabling of TrueType kerning.
- Fix possible crasher with broken fonts with overlapping tables.
- Removed generated files from git again. So, one needs ragel to
bootstrap from the git tree.
API changes:
- hb_shape() and related APIs now abort if buffer direction is
HB_DIRECTION_INVALID. Previously, hb_shape() was calling
hb_buffer_guess_segment_properties() on the buffer before
shaping. The heuristics in that function are fragile. If the
user really wants the old behvaior, they can call that function
right before calling hb_shape() to get the old behavior.
- hb_blob_create_sub_blob() always creates sub-blob with
HB_MEMORY_MODE_READONLY. See comments for the reason.
Overview of changes leading to 0.9.12
Thursday, January 18, 2013
- Build fixes for Sun compiler.
- Minor bug fix.
Overview of changes leading to 0.9.11
Thursday, January 10, 2013
- Build fixes.
- Fix GPOS mark attachment with null Anchor offsets.
- [Indic] Fix old-spec reordering of viramas if sequence ends in one.
- Fix multi-threaded shaper data creation crash.
- Add atomic ops for Solaris.
API changes:
- Rename hb_buffer_clear() to hb_buffer_clear_contents().
Overview of changes leading to 0.9.10
Thursday, January 3, 2013
- [Indic] Fixed rendering of Malayalam dot-reph
- Updated OT language tags.
- Updated graphite2 backend.
- Improved hb_ot_layout_get_size_params() logic.
- Improve hb-shape/hb-view help output.
- Fixed hb-set.h implementation to not crash.
- Fixed various issues with hb_ot_layout_collect_lookups().
- Various build fixes.
New API:
Modified API:
Overview of changes leading to 0.9.9
Wednesday, December 5, 2012
- Fix build on Windows.
- Minor improvements.
Overview of changes leading to 0.9.8
Tuesday, December 4, 2012
- Actually implement hb_shape_plan_get_shaper ().
- Make UCDB data tables const.
- Lots of internal refactoring in OTLayout tables.
- Flesh out hb_ot_layout_lookup_collect_glyphs().
New API:
Overview of changes leading to 0.9.7
Sunday, November 21, 2012
HarfBuzz "All-You-Can-Eat-Sushi" (aka Vancouver) Hackfest and follow-on fixes.
- Fix Arabic contextual joining using pre-context text.
- Fix Sinhala "split matra" mess.
- Fix Khmer shaping with broken fonts.
- Implement Thai "PUA" shaping for old fonts.
- Do NOT route Kharoshthi script through the Indic shaper.
- Disable fallback positioning for Indic and Thai shapers.
- Misc fixes.
hb-shape / hb-view changes:
- Add --text-before and --text-after
- Add --bot / --eot / --preserve-default-ignorables
- hb-shape --output-format=json
New API:
API changes:
- Remove "mask" parameter from hb_buffer_add().
- Rename hb_ot_layout_would_substitute_lookup() and hb_ot_layout_substitute_closure_lookup().
- hb-set.h API const correction.
- Renamed hb_set_min/max() to hb_set_get_min/max().
- Rename hb_ot_layout_feature_get_lookup_indexes() to hb_ot_layout_feature_get_lookups().
- Rename hb_buffer_guess_properties() to hb_buffer_guess_segment_properties().
Overview of changes leading to 0.9.6
Sunday, November 13, 2012
- Don't clear pre-context text if no new context is provided.
- Fix ReverseChainingSubstLookup, which was totally borked.
- Adjust output format of hb-shape a bit.
- Include in-tree. Makes it easier for alternate build systems.
- Fix hb_buffer_set_length(buffer, 0) invalid memory allocation.
- Use ICU LayoutEngine's C API instead of C++. Avoids much headache.
- Drop glyphs for all of Unicode Default_Ignorable characters.
- Misc build fixes.
Arabic shaper:
- Enable 'dlig' and 'mset' features in Arabic shaper.
- Implement 'Phags-pa shaping, improve Mongolian.
Indic shaper:
- Decompose Sinhala split matras the way old HarfBuzz / Pango did.
- Initial support for Consonant Medials.
- Start adding new-style Myanmar shaping.
- Make reph and 'pref' logic introspect the font.
- Route Meetei-Mayek through the Indic shaper.
- Don't apply 'liga' in Indic shaper.
- Improve Malayalam pre-base reordering Ra interaction with Chillus.
Overview of changes leading to 0.9.5
Sunday, October 14, 2012
- Synthetic-GSUB Arabic fallback shaping.
- Misc Indic improvements.
- Add build system support for pthread.
- Imported UCDN for in-tree Unicode callbacks implementation.
- Context-aware Arabic joining.
- Misc other fixes.
- New API:
Overview of changes leading to 0.9.4
Tuesday, Sep 03, 2012
- Indic improvements with old-spec Malayalam.
- Better fallback glyph positioning, specially with Thai / Lao marks.
- Implement dotted-circle insertion.
- Better Arabic fallback shaping / ligation.
- Added ICU LayoutEngine backend for testing. Call it by the 'icu_le' name.
- Misc fixes.
Overview of changes leading to 0.9.3
Friday, Aug 18, 2012
- Fixed fallback mark positioning for left-to-right text.
- Improve mark positioning for the remaining combining classes.
- Unbreak Thai and fallback Arabic shaping.
- Port Arabic shaper to shape-plan caching.
- Use new ICU normalizer functions.
Overview of changes leading to 0.9.2
Friday, Aug 10, 2012
- Over a thousand commits! This is the first major release of HarfBuzz.
- HarfBuzz is feature-complete now! It should be in par, or better, than
both Pango's shapers and old HarfBuzz / Qt shapers.
- New Indic shaper, supporting main Indic scripts, Sinhala, and Khmer.
- Improved Arabic shaper, with fallback Arabic shaping, supporting Arabic,
Sinhala, N'ko, Mongolian, and Mandaic.
- New Thai / Lao shaper.
- Tibetan / Hangul support in the generic shaper.
- Synthetic GDEF support for fonts without a GDEF table.
- Fallback mark positioning for fonts without a GPOS table.
- Unicode normalization shaping heuristic during glyph mapping.
- New experimental Graphite2 backend.
- New Uniscribe backend (primarily for testing).
- New CoreText backend (primarily for testing).
- Major optimization and speedup.
- Test suites and testing infrastructure (work in progress).
- Greatly improved hb-view cmdline tool.
- hb-shape cmdline tool.
- Unicode 6.1 support.
Summary of API changes:
o Changed API:
- Users are expected to only include main header files now (ie. hb.h,
hb-glib.h, hb-ft.h, ...)
- All struct tag names had their initial underscore removed.
Ie. "struct _hb_buffer_t" is "struct hb_buffer_t" now.
- All set_user_data() functions now take a "replace" boolean parameter.
- hb_buffer_create() takes zero arguments now.
Use hb_buffer_pre_allocate() to pre-allocate.
- hb_buffer_add_utf*() now accept -1 for length parameteres,
meaning "nul-terminated".
- hb_direction_t enum values changed.
- All *_from_string() APIs now take a length parameter to allow for
non-nul-terminated strings. A -1 length means "nul-terminated".
- Typedef for hb_language_t changed.
- hb_get_table_func_t renamed to hb_reference_table_func_t.
- hb_ot_layout_table_choose_script()
- Various renames in hb-unicode.h.
o New API:
- hb_buffer_guess_properties()
Automatically called by hb_shape().
- hb_buffer_normalize_glyphs()
- hb_tag_from_string()
- hb-coretext.h
- hb-uniscribe.h
- hb_face_reference_blob()
- hb_face_[sg]et_index()
- hb_face_set_upem()
- hb_font_get_glyph_name_func_t
- hb_font_set_funcs_data()
- hb_ft_font_set_funcs()
- hb_ft_font_get_face()
- hb-gobject.h (work in progress)
- hb_ot_shape_glyphs_closure()
- hb-set.h
- hb_shape_full()
- hb_unicode_combining_class_t
- hb_unicode_compose_func_t
o Removed API:
- hb_ft_get_font_funcs()
- hb_ot_layout_substitute_start()
Overview of changes leading to 0.6.0
Friday, May 27, 2011
- Vertical text support in GPOS
- Almost all API entries have unit tests now, under test/
- All thread-safety issues are fixed
Summary of API changes follows.
* Simple Types API:
o New API:
o Renamed API:
hb_category_t renamed to hb_unicode_general_category_t
o Changed API:
hb_language_t is a typed pointers now
o Removed API:
* Use ISO 15924 tags for hb_script_t:
o New API:
o Changed API:
HB_SCRIPT_* enum members changed value.
* Buffer API streamlined:
o New API:
o Renamed API:
hb_buffer_ensure() renamed to hb_buffer_pre_allocate()
hb_buffer_add_glyph() renamed to hb_buffer_add()
o Removed API:
o Changed API:
hb_buffer_get_glyph_infos() takes an out length parameter now
hb_buffer_get_glyph_positions() takes an out length parameter now
* Blob API streamlined:
o New API:
o Renamed API:
hb_blob_create_empty() renamed to hb_blob_get_empty()
o Removed API:
o Changed API:
hb_blob_create() takes user_data before destroy now
* Unicode functions API:
o Unicode function vectors can subclass other unicode function vectors now.
Unimplemented callbacks in the subclass automatically chainup to the parent.
o All hb_unicode_funcs_t callbacks take a user_data now. Their setters
take a user_data and its respective destroy callback.
o New API:
o Changed API:
hb_unicode_funcs_create() now takes a parent_funcs.
o Removed func getter functions:
* Face API:
o Renamed API:
hb_face_get_table() renamed to hb_face_reference_table()
hb_face_create_for_data() renamed to hb_face_create()
o Changed API:
hb_face_create_for_tables() takes user_data before destroy now
hb_face_reference_table() returns empty blob instead of NULL
hb_get_table_func_t accepts the face as first parameter now
* Font API:
o Fonts can subclass other fonts now. Unimplemented callbacks in the
subclass automatically chainup to the parent. When chaining up,
scale is adjusted if the parent font has a different scale.
o All hb_font_funcs_t callbacks take a user_data now. Their setters
take a user_data and its respective destroy callback.
o New API:
o Removed API:
o Removed func getter functions:
o Changed API:
hb_font_create() takes a face and references it now
hb_font_set_funcs() takes user_data before destroy now
hb_font_set_scale() accepts signed integers now
hb_font_get_contour_point_func_t now takes glyph first, then point_index
hb_font_get_glyph_func_t returns a success boolean now
* Changed object model:
o All object types have a _get_empty() now:
o Added _set_user_data() and _get_user_data() for all object types:
o Removed the _get_reference_count() from all object types:
o Added _make_immutable() and _is_immutable() for all object types except for buffer:
* Changed API for vertical text support
o The following callbacks where removed:
o The following new callbacks added instead:
o The following API removed as such:
o New API added instead:
o The following higher-leve API added for convenience:
* OpenType Layout API:
o New API:
* Glue code:
o New API:
* Version API added:
o New API: