Subversion Repositories Kolibri OS

2010-02-13  Werner Lemberg  

	* Version 2.3.12 released.

	==========================

	Tag sources with `VER-2-3-12'.

	* docs/CHANGES: Updated.

	* docs/VERSION.DLL: Update documentation and bump version number to

	2.3.12.

	* README, Jamfile (RefDoc),

	builds/win32/vc2005/freetype.vcproj, builds/win32/vc2005/index.html,

	builds/win32/vc2008/freetype.vcproj, builds/win32/vc2008/index.html,

	builds/win32/visualc/freetype.dsp,

	builds/win32/visualc/freetype.vcproj,

	builds/win32/visualc/index.html, builds/win32/visualce/freetype.dsp,

	builds/win32/visualce/freetype.vcproj,

	builds/win32/visualce/index.html,

	builds/wince/vc2005-ce/freetype.vcproj,

	builds/wince/vc2005-ce/index.html,

	builds/wince/vc2008-ce/freetype.vcproj,

	builds/wince/vc2008-ce/index.html: s/2.3.11/2.3.12/, s/2311/2312/.

	* include/freetype/freetype.h (FREETYPE_PATCH): Set to 12.

	* builds/unix/configure.raw (version_info): Set to 10:0:4.

2010-02-12  suzuki toshiya  

	Improve autotool version checking to work with beta releases.

	* autogen.sh (check_tool_version): Improve the extraction of version

	number from "tool --version" output.  Some beta releases of

	autotools have extra strings before version number.

2010-02-12  suzuki toshiya  

	Fix overallocating bug in FT_Outline_New_Internal().

	* src/base/ftoutln.c (FT_Outline_New_Internal): The length of

	FT_Outline->points[] should be numPoints, not 2 * numPoints.

	Found by Paul Messmer, see

	http://lists.gnu.org/archive/html/freetype-devel/2010-02/msg00003.html

2010-02-10  Ken Sharp  

	Really fix Savannah bug #28678 (part 2).

	Since we consider `sbw' for the horizontal direction only, we still have

	to synthesize vertical metrics if the user wants to use the vertical

	writing direction.

	* src/cff/cffgload.c (cff_slot_load), src/cid/cidgload.c

	(cid_slot_load_glyph), src/type1/t1gload.c (T1_Load_Glyph):

	Synthesize vertical metrics (only) if FT_LOAD_VERTICAL_LAYOUT is

	set.

2010-02-10  Ken Sharp  

	Really fix Savannah bug #28678 (part 1).

	After long discussion, we now consider the character width vector

	(wx,wy) returned by the `sbw' Type 1 operator as being part of *one*

	direction only.  For example, if you are using the horizontal

	writing direction, you get the horizontal and vertical components of

	the advance width for this direction.  Note that OpenType and CFF fonts

	don't have such a vertical component; instead, the GPOS table can be

	used to generate two-dimensional advance widths (but this isn't

	handled by FreeType).

	* include/freetype/ftincrem.h (FT_Incremental_MetricsRec): Add

	`advance_v' field to hold the vertical component of the advance

	value.

	* src/truetype/ttgload.c (tt_get_metrics), src/cff/cffgload.c

	(cff_slot_load), src/type1/t1gload.c

	(T1_Parse_Glyph_And_Get_Char_String), src/cid/cidgload.c

	(cid_load_glyph): Use it.

2010-02-08  Werner Lemberg  

	* devel/ftoption.h [FT_CONFIG_OPTION_PIC]: Define.

2010-02-04  suzuki toshiya  

	Prevent NULL pointer dereference passed to FT_Module_Requester.

	* src/sfnt/sfdriver.c (sfnt_get_interface): Don't use `module'.

	* src/psnames/psmodule.c (psnames_get_interface): Ditto.

	* src/cff/cffdrivr.c (cff_get_interface): Check NULL `driver'.

	* src/truetype/ttdriver.c (tt_get_interface): Ditto.

2010-01-29  suzuki toshiya  

	Fix memory leaks in previous patch.

	* src/sfnt/sfobjs.c (sfnt_load_face): Don't overwrite the strings

	allocated for face->root.family_name and style_name.

2010-01-29  suzuki toshiya  

	New parameters for FT_Open_Face() to ignore preferred family names.

	Preferred family names should be used for legacy systems that

	can hold only a few faces (<= 4) for a family name. Suggested by

	Andreas Heinrich.

	http://lists.gnu.org/archive/html/freetype/2010-01/msg00001.html

	* include/freetype/ftsnames.h (FT_PARAM_TAG_IGNORE_PREFERRED_FAMILY,

	FT_PARAM_TAG_IGNORE_PREFERRED_SUBFAMILY): Define.

	* src/sfnt/sfobjs.c (sfnt_load_face): Check the arguments and

	ignore preferred family and subfamily names if requested.

2010-01-27  Ken Sharp  

	Fix Savannah bug #28678.

	* src/cff/cffgload.c (cff_slot_load), src/cid/cidgload.c

	(cid_load_glyph): Handle vertical metrics correctly.

	* src/type1/t1gload.c (T1_Parse_Glyph_And_Get_Char_String): Handle

	vertical metrics correctly.

	(T1_Load_Glyph): Don't synthesize vertical metrics.

2010-01-14  Werner Lemberg  

	Make FT_Set_Transform work if no renderer is available.

	* src/base/ftobjs.c (FT_Load_Glyph): Apply `standard' transformation

	if no renderer is compiled into the library.

2010-01-14  Werner Lemberg  

	Fix compilation warning.

	* src/base/ftbase.h: s/LOCAL_DEF/LOCAL/.

	* src/base/ftobjc.s: Include ftbase.h conditionally.

2010-01-11  Kwang Yul Seo  

	Provide inline assembly code for RVCT compiler.

	This is Savannah patch #7059.

	* include/freetype/config/ftconfig.h (FT_MULFIX_ASSEMBLER,

	FT_MulFix_arm) [__CC_ARM || __ARM_CC]: Define.

2010-01-08  Ken Sharp  

	Fix Savannah bug #28521.

	Issue #28226 involved a work-around for a font which used the

	`setcurrentpoint' operator in an invalid way; this operator is only

	supposed to be used with the result of OtherSubrs, and the font used

	it directly.  The supplied patch removed the block of code which

	checked this usage entirely.

	This turns out to be a Bad Thing.  If `setcurrentpoint' is being

	used correctly it should reset the flex flag in the decoder.  If we

	don't do this then the flag never gets reset and we omit any further

	contours from the glyph (at least until we close the path or

	similar).

	* src/psaux/t1decode.c (t1_decoder_parse_charstrings)

	: Handle `flex_state' correctly.

2010-01-05  Werner Lemberg  

	Apply reports from clang static analyzer.

	* src/lzw/ftlzw.c (ft_lzw_file_init), src/base/ftstroke.c

	(FT_Stroker_ParseOutline), src/base/ftsynth.c

	(FT_GlyphSlot_Embolden): Remove dead code.

	* src/base/ftpatent.c (_tt_check_patents_in_table): Initialize

	`offset_i' and `length_i'.

2010-01-05  Ralph Giles  

	Enable the incremental font interface by default.

	Ghostscript requires the incremental font interface for handling

	some Postscript documents.  It is moving to using FreeType as its

	primary renderer; supporting this in the default build makes it

	Ghostscript to be linked against the system FreeType when one is

	available.

	* include/freetype/config/ftoption.h (FT_CONFIG_OPTION_INCREMENTAL):

	Uncomment.

2010-01-05  Werner Lemberg  

	Fix Savannah bug #28395.

	* src/truetype/ttdriver.c (Load_Glyph), src/type1/t1gload.c

	(T1_Loada_Glyph): Don't check `num_glyphs' if incremental interface

	is used.

2010-01-05  Ken Sharp  

	Make Type 1 `seac' operator work with incremental interface.

	This fixes Savannah bug #28480.

	* src/psaux/t1decode.c (t1operator_seac): Don't check `glyph_names'

	if incremental interface is used.

2010-01-04  Ken Sharp  

	Make incremental interface work with TrueType fonts.

	This fixes Savannah bug #28478.

	* src/truetype/ttgload.c (load_truetype_glyph): Don't check

	`glyf_offset' if incremental interface is used.

2009-12-31  Lars Abrahamsson  

	Make compilation with FT_CONFIG_OPTION_PIC work again.

	* src/base/ftglyph.c (FT_Glyph_To_Bitmap) [FT_CONFIG_OPTION_PIC]:

	Declare `library' for FT_BITMAP_GLYPH_CLASS_GET.

	* src/base/ftinit.c (ft_destroy_default_module_classes,

	ft_create_default_module_classes): Use proper casts (needed for C++

	compilation).

	* src/sfnt/ttcmap.c (tt_cmap13_class_rec): Use FT_DEFINE_TT_CMAP.

2009-12-22  Marc Kleine-Budde  

	Make freetype-config aware of $SYSROOT.

	This is Savannah patch #7040.

	* builds/unix/freetype-config.in: Decorate with ${SYSROOT} where

	appropriate.

2009-12-20  Werner Lemberg  

	Fix compiler warning.

	Reported by Sean.

	* src/base/ftdbgmem.c [!FT_DEBUG_MEMORY]: ANSI C doesn't like empty

	source files; however, some compilers warn about an unused variable

	declaration.  This is now replaced with a typedef.

2009-12-18  Werner Lemberg  

	Fix Savannah bug #28320.

	There exist corrupt, subsetted fonts (embedded in PDF files) which

	contain a private dict that ends with an unterminated floating point

	number (no operator following).  We now ignore this error (as

	acrobat does).

	* src/cff/cffparse.c (cff_parser_run): Don't emit a syntax error for

	unterminated floating point numbers.

2009-12-16  Werner Lemberg  

	Really fix compiler warnings.

	Reported by Sean.

	* src/truetype/ttgxvar.c (GX_PT_POINTS_ARE_WORDS,

	GX_PT_POINT_RUN_COUNT_MASK): Convert enum values to macros.

2009-12-16  suzuki toshiya  

	Improve configure.raw to copy some options from CFLAGS to LDFLAGS.

	The linker of Mac OS X 10.6 is sensitive to the architecture.  If

	the architectures are specified explicitly for the C compiler, the

	linker requires the architecture specifications too.

	* builds/unix/configure.raw: Replace `-isysroot' option parser by

	more generic argument parser.

2009-12-15  Werner Lemberg  

	Fix compiler warnings.

	Reported by Sean.

	* src/truetype/ttgxvar.c (ft_var_readpackeddeltas): Fix counter data

	type.

2009-12-14  Ken Sharp  

	Ignore invalid `setcurrentpoint' operations in Type 1 fonts.

	This fixes Savannah bug #28226.

	At least two wild PostScript files of unknown provenance contain

	Type 1 fonts, apparently converted from TrueType fonts in earlier

	PDF versions of the files, which use the `setcurrentpoint' operator

	inappropriately.

	FreeType currently throws an error in this case, but Ghostscript and

	Adobe Distiller both accept the fonts and ignore the problem.  This

	commit #ifdefs out the check so PostScript interpreters using

	FreeType can render these files.

	The specification says `setcurrentpoint' should only be used to set

	the point after a `Subr' call, but these fonts use it to set the

	initial point to (0,0).  Unnecessarily so, as they correctly use an

	`hsbw' operation which implicitly sets the initial point.

	* src/psaux/t1decode.c (t1_decoder_parse_charstrings)

	: Comment out code.

2009-12-14  Bram Tassyns  

	Fix parsing of /CIDFontVersion.

	This fixes Savannah bug #28287.

	* src/cid/cidtoken.h: `cid_version' in CID_FaceInfoRec (in

	t1tables.h) is of type FT_Fixed.

2009-12-14  Werner Lemberg  

	Trace glyph index in CID module.

	Suggested in Savannah patch #7023.

	* src/cid/cidgload.c (cid_load_glyph): Add tracing message.

2009-12-03  Werner Lemberg  

	Fix compiler warnings.

	* src/truetype/ttgload.c (tt_get_metrics): Put `Exit' label into the

	proper preprocessor conditional.

	* src/pfr/pfrobjs.c (pfr_slot_load): Pacify gcc.

2009-11-25  John Tytgat  

	Better handling of start of `eexec' section.

	This fixes Savannah bug #28090.

	* src/type1/t1parse.c (T1_Get_Private_Dict): Skip all whitespace

	characters before start of `eexec' section.

2009-11-20  Werner Lemberg  

	Fix Savannah bug #27742.

	* src/base/ftstroke.c (ft_stroker_outside): Avoid silent division by

	zero, using a threshold for `theta'.

2009-11-20  Werner Lemberg  

	Fix Savannah bug #28036.

	* src/type1/t1afm.c (t1_get_index): Fix comparison.

2009-11-16  Werner Lemberg  

	Fix compiler warnings.

	Reported by Kevin Blenkinsopp .

	* src/sfnt/ttload.c (check_table_dir): Use proper data type.

2009-11-15  Werner Lemberg  

	Really fix FreeDesktop bug #21197.

	This also fixes Savannah bug #28021.

	* src/autofit/aflatin.c (af_latin_metrics_check_digits),

	src/autofit/aflatin2.c (af_latin2_metrics_check_digits): Fix loop.

2009-11-15  Werner Lemberg  

	Add tracing messages for advance values.

	* src/base/ftobjs.c (FT_Load_Glyph), src/truetype/ttgload.c

	(TT_Get_HMetrics, TT_Get_VMetrics): Do it.

2009-11-08  Werner Lemberg  

	Fix compiler warning.

	Reported by Jeremy Manson .

	* src/truetype/ttgload.c (load_truetype_glyph): Initialize `error'.

2009-11-04  Werner Lemberg  

	Remove compiler warning.

	Reported by Sean McBride .

	* src/tools/apinames.c (read_header_file): Use a cast to

	`int', as specified in the printf(3) man page.

2009-11-04  Werner Lemberg  

	Fix Savannah bug #27921.

	* src/cff/cffobjs.c (cff_face_init), src/cid/cidobjs.c

	(cid_face_init), src/type1/t1afm.c (T1_Read_Metrics),

	src/type1/t1objs.c (T1_Face_Init): Don't use unsigned constant

	values for rounding if the argument can be negative.

2009-11-03  Bram Tassyns  

	Add basic support for Type1 charstrings in CFF.

	This fixes Savannah bug #27922.

	* src/cff/cffgload.c (CFF_Operator, cff_argument_counts): Handle

	`seac', `sbw', and `setcurrentpoint' opcodes.

	(cff_compute_bias): Add parameter to indicate the charstring type.

	Update all callers.

	(cff_operator_seac): Add parameter for side bearing.

	(cff_decoder_parse_charstrings): Updated for more Type1 support.

2009-11-03  Werner Lemberg  

	Return correct `linearHoriAdvance' value for embedded TT bitmaps too.

	Reported by Jeremy Manson .

	src/truetype/ttgload.c (load_truetype_glyph): Add parameter to

	quickly load the glyph header only.

	Update all callers.

	(tt_loader_init): Add parameter to quickly load the `glyf' table

	only.

	Update all callers.

	(TT_Load_Glyph): Compute linear advance values for embedded bitmap

	glyphs too.

2009-11-03  Werner Lemberg  

	Improve code readability.

	* src/ttgload.c (load_truetype_glyph): Move metrics calculation

	to...

	(tt_get_metrics): This new function.

2009-10-26  Bram Tassyns  

	Fix Savannah bug #27811.

	* src/truetype/ttxgvar.c (ft_var_readpackeddeltas): Fix

	signed/unsigned mismatch.

2009-10-19  Ning Dong  

	Fix handling of `get' and `put' CFF instructions.

	* src/cff/cffgload.c (cff_decoder_parse_charstrings) 

	cff_op_put>: Appendix B of Adobe Technote #5177 limits the number of

	elements for the `get' and `put' operators to 32.

	* src/cff/cffgload.h (CFF_MAX_TRANS_ELEMENTS): Define.

	(CFF_Decoder): Use it for `buildchar' and remove `len_buildchar'.

2009-10-18  Werner Lemberg  

	Fix handling of `dup' CFF instruction.

	Problem and solution reported by Ning Dong .

	* src/cff/cffgload.c (cff_decoder_parse_charstrings) :

	Increase `args' by 2, not 1.

2009-10-10  Werner Lemberg  

	* Version 2.3.11 released.

	==========================

	Tag sources with `VER-2-3-11'.

	* docs/VERSION.DLL: Update documentation and bump version number to

	2.3.11.

	* README, Jamfile (RefDoc), builds/win32/visualc/index.html,

	builds/win32/visualc/freetype.dsp,

	builds/win32/visualc/freetype.vcproj,

	builds/win32/visualce/index.html,

	builds/win32/visualce/freetype.dsp,

	builds/win32/visualce/freetype.vcproj: s/2.3.10/2.3.11/, s/2310/2311/.

	* include/freetype/freetype.h (FREETYPE_PATCH): Set to 11.

	* builds/unix/configure.raw (version_info): Set to 9:22:3.

2009-10-10  Werner Lemberg  

	* docs/CHANGES, docs/release: Updated.

2009-10-10  suzuki toshiya  

	* src/pcf/pcfread.c (pcf_get_properties): Fix a bug in the nprops

	truncation.  Reported by Martin von Gagern and Peter Volkov.

	https://bugs.gentoo.org/288357 and https://bugs.gentoo.org/288256

2009-10-06  Werner Lemberg  

	* Version 2.3.10 released.

	==========================

	Tag sources with `VER-2-3-10'.

	* builds/toplevel.mk (major, minor, patch): Fix regexp to allow more

	than a single digit.

	(dist): We now use git.

	* docs/VERSION.DLL: Update documentation and bump version number to

	2.3.10.

	* README, Jamfile (RefDoc), builds/win32/visualc/index.html,

	builds/win32/visualc/freetype.dsp,

	builds/win32/visualc/freetype.vcproj,

	builds/win32/visualce/index.html,

	builds/win32/visualce/freetype.dsp,

	builds/win32/visualce/freetype.vcproj: s/2.3.9/2.3.10/, s/239/2310/.

	* include/freetype/freetype.h (FREETYPE_PATCH): Set to 10.

	* builds/unix/configure.raw (version_info): Set to 9:21:3.

2009-10-06  Werner Lemberg  

	Fix `make multi'.

	* src/cache/ftccache.c, src/cache/ftcsbits.c (FT_COMPONENT): Define.

	* src/sfnt/sfdriver.c: Include FT_INTERNAL_DEBUG_H.

2009-09-27  suzuki toshiya  

	[cache] Fix Savannah bug #27441, clean up Redhat bugzilla #513582.

	Tricky casts in FTC_{CACHE,GCACHE,MRULIST}_LOOKUP_CMP() are removed.

	Now these functions should be called with FTC_Node or FTC_MruNode

	variable, and the caller should cast them to appropriate pointers to

	concrete data.  These tricky casts can GCC-4.4 optimizer (-O2)

	confused and the crashing binaries are generated.

	* src/cache/ftcmru.h (FTC_MRULIST_LOOKUP_CMP): Drop tricky cast.

	Now the 4th argument `node' of this function should be typed as

	FTC_MruNode.

	* src/cache/ftcglyph.h (FTC_GCACHE_LOOKUP_CMP): For inline

	implementation, new temporal variable FTC_MruNode `_mrunode' to take

	the pointer from FTC_MRULIST_LOOKUP_CMP().  For non-inline

	implementation, tricky cast is dropped.

	* src/cache/ftcmanag.c (FTC_SIZE_NODE): New macro casting

	to FTC_SizeNode.

	(FTC_Manager_LookupSize): Replace FTC_SizeNode `node' by FTC_MruNode

	`mrunode', and FTC_SIZE_NODE() is inserted.

	(FTC_FACE_NODE): New macro casting to FTC_FaceNode.

	(FTC_Manager_LookupFace) Replace FTC_FaceNode `node' by FTC_MruNode

	`mrunode', and FTC_FACE_NODE() is inserted.

	* src/cache/ftcbasic.c (FTC_ImageCache_Lookup): Change the type of

	`node' from FTC_INode to FTC_Node.  Extra casting macro FTC_NODE()

	is dropped.

	(FTC_ImageCache_LookupScaler): Ditto.

	(FTC_SBitCache_Lookup): Change the type of `node' from FTC_SNode to

	FTC_Node.  Extra casting macro FTC_NODE() is dropped.  FTC_SNODE()

	is inserted.

	(FTC_SBitCache_LookupScaler): Ditto.

	* src/cache/ftccmap.c (FTC_CMapCache_Lookup): Change the type of

	`node' from FTC_CMapNode to FTC_Node.  Extra casting macro

	FTC_NODE() is dropped, FTC_CMAP_NODE() is inserted.

2009-09-25  suzuki toshiya  

	[cache, psaux, type1] Fix for multi build.

	In multi build, some cpp functions are left as unresolved symbols.

	* src/cache/ftcbasic.c: Include FT_INTERNAL_DEBUG_H for FT_TRACE1().

	* src/psaux/t1decode.c: Include FT_INTERNAL_CALC_H for

	FIXED_TO_INT().

	* src/type1/t1gload.c: Ditto.

	* src/type1/t1objs.c: Ditto.

2009-09-25  suzuki toshiya  

	[autofit] Fix for multi build.

	* src/autofit/afmodule.h: Include FT_INTERNAL_OBJECTS_H to use

	FT_DECLARE_MODULE() macro in multi build.

	* src/autofit/aflatin.c: Include  to handle

	FT_ADVANCES_H correctly in multi build.

2009-09-24  suzuki toshiya  

	[cache] Check the face filled by FTC_Manager_LookupFace().

	* src/cache/ftcbasic.c (ftc_basic_family_get_count): Return

	immediately if FTC_Manager_LookupFace() fills face by NULL.  Such

	case can occur when the code is optimized by GCC-4.2.x.

2009-09-23  Werner Lemberg  

	* docs/CHANGES: Updated.

2009-09-12  Werner Lemberg  

	[raster] Fix 5-levels grayscale output.

	This was broken since version 2.3.0.

	* src/raster/ftraster.c (count_table): Use pre-2.3.0 values (which

	were then computed dynamically).

	(Vertical_Gray_Sweep_Step): Updated.

	(ft_black_render): Initialize `worker->gray_lines' (problem found by

	valgrind).

	(FT_RASTER_OPTION_ANTI_ALIASING, DEBUG_RASTER): Dont' #undef, just

	comment out.

2009-09-12  suzuki toshiya  

	Improve configure.raw for cross build.

	* builds/unix/configure.raw: Remove temporal files created by the

	suffix checking for CC_BUILD.  Set XX_ANSIFLAGS and XX_CFLAGS when

	cross compiler is GCC.  AC_PROG_CC checks whether the cross compiler

	is GCC, its result is stored in GCC.

2009-09-12  suzuki toshiya  

	[BDF] Modify hash API to take size_t value instead of void *.

	The hash API in BDF driver is designed to be generic, it takes

	void * typed data.  But BDF driver always gives an unsigned long

	integer (the index to a property).  To reduce non-essential

	casts from unsigned long to void* and from void* to unsigned

	long, the hash API is changed to take size_t integer.

	The issue of incompatible cast between unsigned long and void*

	on LLP64 platform is reported by NightStrike from MinGW-Win64

	project.  See

	http://lists.gnu.org/archive/html/freetype/2009-09/msg00000.html

	* src/bdf/bdf.h: The type of hashnode->data is changed from

	void* to size_t.

	* src/bdf/bdflib.c (hash_insert): Get size_t data, instead of

	void* data.

	(bdf_create_property): Get the name length of new property by

	size_t variable, with a cut-off at FT_ULONG_MAX.

	(_bdf_set_default_spacing): Get the name length of the face by

	size_t variable, with a cut-off at 256.

	(bdf_get_property): Get the property id by size_t variable to

	reduce the casts between 32-bit prop ID & hashnode->data during

	simple copying.

	(_bdf_add_property): Ditto.

	(_bdf_parse_start): Calculate the index to the property array

	by size_t variable.

	(bdf_get_font_property): Drop a cast to unsigned long.

2009-09-10  suzuki toshiya  

	[Win64] Improve the computation of random seed from stack address.

	On LLP64 platform, the conversion from pointer to FT_Fixed need

	to drop higher 32-bit.  Explict casts are required. Reported by

	NightStrike from MinGW-w64 project.  See

	http://lists.gnu.org/archive/html/freetype/2009-09/msg00000.html

	* src/cff/cffgload.c: Convert the pointers to FT_Fixed explicitly.

	* src/psaux/t1decode.c: Ditto.

2009-09-03  Werner Lemberg  

	[raster] Improvements for stand-alone mode.

	* src/raster/rules.mk: Don't handle ftmisc.h.  It is needed for

	stand-alone mode only.

	* src/raster/ftmisc.h (FT_MemoryRec , FT_Alloc_Func, FT_Free_Func,

	FT_Realloc_Func): Copy declarations from ftsystem.h.

2009-09-02  Bram Tassyns  

	Improve vertical metrics calculation (Savannah bug #27364).

	The calculation of `vertBearingX' is not defined in the OTF font

	spec so FreeType does a `best effort' attempt.  However, this value

	is defined in the PDF and PostScript specs, and that algorithm is

	better than the one FreeType currently uses:

	  FreeType: Use the middle of the bounding box as the X coordinate

	            of the vertical origin.

	  Adobe PDF spec: Use the middle of the horizontal advance vector as

	                  the X coordinate of the vertical origin.

	FreeType's algorithm goes wrong if you have a really small glyph

	(like the full-width, circle-like dot at the end of the sentence, as

	used in CJK scripts) with large bearings.  With the FreeType

	algorithm this dot gets centered on the baseline; with the PDF

	algorithm it gets the correct location (in the top right).  Note

	that this is a serious issue, it's like printing the dot at the end

	of a Roman sentence at the center of the textline instead of on the

	baseline like it should. So i believe the PDF spec's algorithm

	should be used in FreeType as well.

	The `vertBearingY' value for such small glyphs is also very strange

	if no `vmtx' information is present, since the height of the bbox is

	not representable for the height of the glyph visually (the

	whitespace up to the baseline is part of the glyph).  The fix also

	includes some code for a better estimate of `vertBearingY'.

	* src/base/ftobjs.c (ft_synthesize_vertical_metrics): `vertBearingX'

	is now calculated as described by the Adobe PDF Spec.  Estimate for

	`vertBearingY' now works better for small glyphs completely above or

	below the baseline into account.

	* src/cff/cffgload.c (cff_slot_load): `vertBearingX' is now

	calculated as described by the Adobe PDF Spec.  Vertical metrics

	information was always ignored when FT_CONFIG_OPTION_OLD_INTERNALS

	was not defined.

	* src/truetype/ttgload.c (compute_glyph_metrics): `vertBearingX' is

	now calculated as described by the Adobe PDF Spec.

2009-09-01  John Tytgat  

	Fix custom cmap for empty Type 1 font (Savannah bug #27294).

	* include/freetype/internal/t1types.h (T1_EncodingRecRec_): Update

	comment to reflect revised code_last meaning.

	* src/type1/t1load.c (T1_Open_Face), src/type42/t42objs.c

	(T42_Open_Face): Assign max_char as highest character code + 1 and

	use this for T1_EncodingRecRec_::code_last.

	* src/psaux/t1cmap.c (t1_cmap_custom_init): Follow revised

	T1_EncodingRecRec_::code_last meaning.

2009-08-25  Werner Lemberg  

	Fix rendering of horizontally compressed CFFs.

	Bug reported by Ivan Nincic .

	* src/cff/cffgload.c (cff_slot_load): Thinko: Check `xx' element of

	`font_matrix' also.

	* docs/CHANGES: Updated.

2009-08-03  suyu0925@gmail.com

	Don't call `ft_fseek' every time when executing `ft_fread'.

	* src/base/ftstream.c (FT_Stream_Seek), src/base/ftsystem.c

	(ft_ansi_stream_io): Implement it.

2009-07-31  suzuki toshiya  

	sfnt: Cast a charcode to 32-bit in cmap format 14 parser.

	* src/sfnt/ttcmap.c (tt_cmap14_char_var_index,

	tt_cmap14_char_var_isdefault, tt_cmap14_char_variants,

	tt_cmap14_variant_chars): Correct mismatches from

	FT_CMap_CharVarIndexFunc prototype, FT_ULong arguments

	are replaced by FT_UInt32 arguments.

2009-07-31  suzuki toshiya  

	sfnt: Cast a charcode to 32-bit in cmap format 12 parser.

	* src/sfnt/ttcmap.c (tt_cmap12_char_next):

	Insert explicit cast from FT_UFast to FT_UInt32

	for return value.

2009-07-31  suzuki toshiya  

	psaux: Fix a few casts to FT_Int32 value.

	* src/psaux/t1decode.c (t1_decoder_parse_charstrings):

	Fix a few casts setting `value' from FT_Long to FT_Int32,

	because `value' is typed as FT_Int32 since 2009-06-22.

2009-07-31  suzuki toshiya  

	sfnt: Fix a data type mismatching with its source.

	* src/sfnt/ttcmap.c (tt_cmap13_char_next): Fix the

	type of `gindex' from FT_ULong to FT_UInt because

	it is set by FT_UInt tt_cmap13_char_map_binary() or

	TT_CMap13->cur_gindex.

2009-07-31  suzuki toshiya  

	sfnt: Extend a few local variables to load 32-bit values.

	* src/sfnt/ttkern.c (tt_face_load_kern): Extend `count'

	and `kern' to load 32-bit values.

2009-07-31  suzuki toshiya  

	pfr: Extend `num_aux' to take 32-bit value.

	* src/pfr/pfrload.c (pfr_phy_font_load): Extend

	`num_aux' to load 32-bit value.

2009-07-31  suzuki toshiya  

	pcf: Truncate FT_ULong `nprops' to fit to int PCF_Face->nprops.

	* src/pcf/pcfread.c (pcf_get_properties): Load `nprops'

	as FT_ULong value from PCF file, but truncate it as

	int to fit PCF_Face->nprops.  The number of truncated

	properties is shown in the trace message.

2009-07-31  suzuki toshiya  

	gxvalid: Extend a few local variables to reduce the casts.

	* src/gxvalid/gxvmorx.c (gxv_morx_subtables_validate):

	Extend `type' and `rest' to take FT_ULong values.

2009-07-31  suzuki toshiya  

	gxvalid: Extend `settingTable' to take 32-bit offset.

	* src/gxvalid/gxvfeat.c (gxv_feat_name_validate):

	Extend `settingTable' to take 32-bit offset.

2009-07-31  suzuki toshiya  

	autofit: Cast FT_Long glyph_count to compare with FT_UInt GID.

	* src/autofit/afglobal.c (af_face_globals_is_digit,

	af_face_globals_compute_script_coverage): Cast FT_Long

	globals->glyph_count to FT_ULong, to compare with FT_UInt

	gindex.

2009-07-31  suzuki toshiya  

	smooth: Exclude 16-bit system in invalid pitch/height check.

	* src/smooth/ftsmooth.c (ft_smooth_render_generic):

	pitch and height are typed as FT_UInt but checked to fit

	16-bit range, to avoid the overflows.  On 16-bit system,

	this checking inserts a conditional that never occurs.

2009-07-03  suzuki toshiya  

	cff: Type large constants > 0x7FFF as long for 16-bit systems.

	* src/cff/cffload.c (cff_charset_load): Type large

	constants > 0x7FFF as long, because normal constants

	are typed signed integer that is less than 0x8000 on

	16-bit systems.

2009-07-31  suzuki toshiya  

	base: Remove an unused variable.

	* src/base/ftglyph.c (FT_Glyph_To_Bitmap): Remove an

	unused variable `library'.  glyph->library is used.

2009-07-31  suzuki toshiya  

	cache: Check higher bits in flags for non ILP32 systems.

	4 public functions ought to take FT_ULong flags, but take

	FT_UInt flags. To keep binary compatibility, we drop higher

	bits on non ILP32 platforms,

		ILP64 systems: No drop occurs.

		LP64 systems: Higher bits are not used.

		16-bit systems: Drop can occur.

	See

	http://lists.gnu.org/archive/html/freetype-devel/2008-12/msg00065.html

	These functions will be refined to take FT_ULong flags in

	next bump with incompatible API change.

	* src/cache/ftcbasic.c (FTC_ImageCache_Lookup):

	Check `flags' in `type', the 2nd argument.

	(FTC_SBitCache_Lookup): Ditto.

	(FTC_ImageCache_LookupScaler): Check `load_flags',

	the 3rd argument.

	(FTC_SBitCache_LookupScaler): Ditto.

2009-07-31  suzuki toshiya  

	sfnt: Ignore invalid GIDs in glyph name lookup.

	* include/freetype/internal/fttrace.h:

	New trace module for sfdriver.c is added.

	* src/sfnt/sfdriver.c (sfnt_get_name_index):

	Restrict glyph name lookup to FT_UInt GID.

	Genuine TrueType can hold 16-bit glyphs.

2009-07-31  suzuki toshiya  

	pcf: Fix a comparison between FT_Long and FT_ULong.

	* src/pcf/pcfread.c (pcf_get_bitmaps): Return an error

	if PCF_Face->nemetrics is negative.

2009-07-31  suzuki toshiya  

	gxvalid: Guarantee `nFeatureFlags' size up to 32-bit.

	* src/gxvalid/gxvmort.c (gxv_mort_featurearray_validate):

	Extend the 3rd argument `nFeatureFlags' to FT_ULong.

	* src/gxvalid/gxvmort.h: Ditto.

2009-07-31  suzuki toshiya  

	sfnt: Insert explicit cast for LP64 system.

	* src/sfnt/ttkern.c (tt_face_load_kern): Insert

	cast from unsigned long to FT_UInt32.

2009-07-31  suzuki toshiya  

	gxvalid: Guarantee `just' table size upto 32-bit.

	* src/gxvalid/gxvjust.c (gxv_just_validate):

	The type of `offset' is changed from FT_UInt to

	FT_Offset, for 16-bit platforms.

2009-07-31  suzuki toshiya  

	gxvalid: Guarantee `trak' table size upto 32-bit.

	* src/gxvalid/gxvtrak.c (gxv_trak_validate):

	The type of `offset' is changed from FT_UInt to

	FT_Offset, for 16-bit platforms.

2009-07-31  suzuki toshiya  

	type1: Fix a data type mismatching with its source.

	* include/freetype/internal/t1types.h: The type of

	T1_Face->buildchar is matched with T1_Decorder->top.

2009-07-31  suzuki toshiya  

	pfr: Fix a data type mismatching with its source.

	* src/pfr/pfrtypes.h: The type of PFR_KernItem->offset

	is extended from FT_UInt32 to FT_Offset, because it is

	calculated with the pointer difference, in

	pfr_extra_item_load_kerning_pairs().

2009-07-31  suzuki toshiya  

	pfr: Fix a data type mismatching with its source.

	* src/pfr/pfrtypes.h: The type of PFR_PhysFont->chars_offset

	is extended from FT_UInt32 to FT_Offset, because it is

	calculated with the pointer difference in pfr_phy_font_load().

2009-07-31  suzuki toshiya  

	pfr: Fix a data type mismatching with its source.

	* src/pfr/pfrtypes.h: The type of PFR_PhyFont->bct_offset

	is extended from FT_UInt32 to FT_Long, because it is

	loaded by FT_STREAM_POS() in pfr_phy_font_load().

2009-07-31  suzuki toshiya  

	smooth: Improve the format in debug message.

	* src/smooth/ftgrays.c (gray_dump_cells): Improve the

	format specifications to dump variables.

2009-07-31  suzuki toshiya  

	sfnt: Fix a data type mismatching with its source.

	* src/sfnt/sfobjs.c (sfnt_load_face): The type of

	local `flags' is matched with FT_Face->face_flags.

2009-07-31  suzuki toshiya  

	psaux: Fix a data type mismatching with its source.

	* include/freetype/internal/psaux.h: The type of

	T1_DecorderRec.buildchar is matched with

	T1_DecorderRec.top.

2009-07-31  suzuki toshiya  

	truetype: Extend TrueType GX packed deltas to FT_Offset.

	* src/truetype/ttgxvar.c (ft_var_readpackeddeltas):

	The type of 2nd argument `delta_cnt' is changed from

	FT_Int to FT_Offset, because its source can be cvt

	table size calculated from stream position.

2009-07-31  suzuki toshiya  

	truetype: Extend mmvar_len to hold size_t values.

	* src/truetype/ttgxvar.h: The type of

	GX_BlendRec.mmvar_len is changed from FT_Int to

	FT_Offset, because TT_Get_MM_Var() calculates it

	by sizeof() results.

2009-07-31  suzuki toshiya  

	truetype: Check invalid function number in IDEF instruction.

	* src/truetype/ttinterp.c (Ins_IDEF): Check

	if the operand fits to 8-bit opcode limitation.

2009-07-31  suzuki toshiya  

	truetype: Check invalid function number in FDEF instruction.

	* src/truetype/ttinterp.c (Ins_FDEF): Check

	if the operand fits 16-bit function number.

2009-07-31  suzuki toshiya  

	truetype: Truncate the deltas of composite glyph at 16-bit values.

	* src/truetype/ttgload.c (load_truetype_glyph):

	Insert cast from FT_Long (deltas[i].{x,y}) to

	FT_Int16 in the summation of deltas[] for composite

	glyphs.  Because deltas[i] is typed as FT_Pos,

	its component x, y are typed as FT_Long, but

	their sources are always FT_Int16 when they are

	loaded by ft_var_readpackeddeltas().  However,

	the limitation about the summed deltas is unclear.

2009-07-31  suzuki toshiya  

	truetype: Truncate the instructions upto 16-bit per a glyph.

	* src/truetype/ttgload.c (TT_Hint_Glyph): Truncate

	the instructions upto 16-bit length per a glyph.

2009-07-31  suzuki toshiya  

	truetype: Cast the numerical operands to 32-bit for LP64 systems.

	* src/truetype/ttinterp.c (Ins_SPHIX, INS_MIAP,

	Ins_MIRP): Insert cast from long (args[], the

	operands passed to TrueType operator) to FT_Int32

	(the argument of TT_MulFix14()).

2009-07-31  suzuki toshiya  

	truetype: Cast the project vector to 32-bit for LP64 system.

	* src/truetype/ttinterp.c (Project, DualProject):

	Insert casts from FT_Pos (the arguments `dx', `dy')

	to FT_UInt32 (the argument to TT_DotFix14()).

2009-07-31  suzuki toshiya  

	truetype: Cast the scaling params to 32-bit for LP64 system.

	* src/truetype/ttgload.c (TT_Process_Composite_Component):

	Insert casts from long (return value of FT_MulFix()) to

	FT_Int32 (the argument to FT_SqrtFixed()).

2009-07-31  suzuki toshiya  

	sfnt: Cast a character code to FT_UInt32 for LP64 system.

	* src/sfnt/ttcmap.c (tt_cmap14_char_map_nondef_binary,

	tt_cmap14_variants, tt_cmap14_char_variants,

	tt_cmap14_def_char_count, tt_cmap14_get_def_chars,

	tt_cmap14_get_nondef_chars, tt_cmap14_variant_chars)

	Insert casts when FT_UInt32 variable is loaded by

	TT_NEXT_{UINT24|ULONG}.  Because most of them are

	compared with FT_UInt32 values in public API, replacing

	FT_UFast is not recommended.

2009-07-31  suzuki toshiya  

	sfnt: Cast a character code to FT_UInt32 for LP64 system.

	* src/sfnt/ttcmap.c (tt_cmap4_init, tt_cmap4_next):

	Insert the casts from unsigned long constant to

	FT_UInt32.

2009-07-31  suzuki toshiya  

	sfnt: Extend TT_BDF->strings_size to FT_ULong for huge BDF.

	* include/freetype/internal/tttypes.h: The type

	of TT_BDF->string_size is extended from FT_UInt32

	to FT_ULong, because BDF specification does not

	restrict the length of string.

	* src/sfnt/ttbdf.c: The scratch variable `strings'

	to load TT_BDF->string_size is matched with

	TT_BDF->string_size.

2009-07-31  suzuki toshiya  

	psaux: Handle the string length by FT_Offset variables.

	* src/psaux/afmparse.c (afm_parser_next_key,

	afm_tokenize, afm_parse_track_kern,

	afm_parse_kern_pairs, afm_parse_kern_data,

	afm_parser_skip_section, afm_parser_parse):

	The length of key is handled by FT_Offset,

	instead of FT_UInt.  Although the length of

	PostScript strings or name object is 16-bit,

	AFM_STREAM_KEY_LEN() calculates the length

	from the pointer difference.

	* src/psaux/afmparse.h (afm_parser_next_key):

	Ditto.

2009-07-31  suzuki toshiya  

	pcf: Fix some data types mismatching with their sources.

	* src/pcf/pcfread.c (pcf_get_bitmaps): The types

	of `nbitmaps', `i', `sizebitmaps' are matched with