]> code.ossystems Code Review - openembedded-core.git/commitdiff
harfbuzz: update 1.2.1 -> 1.2.3
authorAndre McCurdy <armccurdy@gmail.com>
Fri, 26 Feb 2016 21:12:18 +0000 (13:12 -0800)
committerRichard Purdie <richard.purdie@linuxfoundation.org>
Wed, 2 Mar 2016 22:39:05 +0000 (22:39 +0000)
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.

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
meta/recipes-graphics/harfbuzz/harfbuzz_1.2.3.bb [moved from meta/recipes-graphics/harfbuzz/harfbuzz_1.2.1.bb with 88% similarity]

similarity index 88%
rename from meta/recipes-graphics/harfbuzz/harfbuzz_1.2.1.bb
rename to meta/recipes-graphics/harfbuzz/harfbuzz_1.2.3.bb
index fead2d379401b3de2f4e0075ffedcf46515c1c12..cfe885fe7e1289b198f2c7a6a2ee61c95c5ab791 100644 (file)
@@ -12,8 +12,8 @@ DEPENDS = "glib-2.0 cairo fontconfig freetype"
 
 SRC_URI = "http://www.freedesktop.org/software/harfbuzz/release/${BP}.tar.bz2"
 
-SRC_URI[md5sum] = "c50034739e4b5ea85d46199bf7859ea6"
-SRC_URI[sha256sum] = "0c189aa386c5ea8d7c315adf273f56f48f412081a367b3f92abc2d0855fa2226"
+SRC_URI[md5sum] = "da8d97e262e6ef8288e1ae76369421bd"
+SRC_URI[sha256sum] = "8216d2404aaab7fde87be0365a90d64aa6c55928e104557cfcb37b54a096cb8c"
 
 inherit autotools pkgconfig lib_package