[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Devel] Latest changes
From: |
David Turner |
Subject: |
[Devel] Latest changes |
Date: |
Thu, 07 Nov 2002 12:35:14 +0100 |
User-agent: |
Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.2a) Gecko/20020910 |
Hello,
I've made several relatively important changes last night:
- hacked our copy of the zlib sources to prevent it
from exporting any public symbols. This allows you
to link FreeType and zlib with the same program on
all platforms.
note that external dependencies on "calloc" and "free"
have also been removed, so the ftgzip component is now
really "embedded"
- add support for FT_CONFIG_OPTION_SYSTEM_ZLIB in
"ftoption.h" and "ftgzip.c". This has not been tested
but should allow linkage against system-wide zlib
installations.
The hard part is of course getting all the build
systems to properly configure the "feature".
In the meantime, the current solution now works on
all platforms and should not cause any problems
(let me know if you find them)
- disabled serif stem widths quantization in the auto
hinter. This produces slightly better results for
certain fonts.
- added a new FT_RENDER_MODE_LIGHT corresponding to
"smoother" anti-aliased hinting, i.e. avoiding
stem width quantization, but still performing
stem grid alignment. This in order to match the
"best contrast/best shapes" selection found in
RedHat 8.0 (as well as recent Gnome 2 snapshots)
with FT_RENDER_MODE_DEFAULT and FT_RENDER_MODE_LIGHT,
respectively.
*NOTE* : this is not implemented properly - probably
tonight -
- removed the hideous "variable 'p' might be clobbered
by longjmp() warning". Fact was that this was not a
compiler bug after all. There *is* a difference between
a pointer to a volatile variable, and a volatile pointer
to a variable, so the fix was to change:
volatile const FT_Byte* p;
into:
const FT_Byte* volatile p;
silly me :-)
- removed a small hack in the small bitmap cache that caused
some problems with certain fonts containing embedded bitmaps
(e.g. BATANG.TTC). This created double frees of memory
which turned to generate segmentation faults sometimes !!
The hack was a way to avoid memory copies by re-using bitmap
buffers, but I don't think removing it should have a significant
impact on performance.
- started completing the code in "src/otlayout" by adding support
for the first GSUB lookup types. Don't hold your breath, this
doesn't compile yet :-)
I hope to complete the FT_RENDER_MODE_LIGHT hinting mode tonight,
and 2.1.3rc5 should be out. I don't intend to make _any_ changes
to it before 2.1.3 release, except major bug fixes.
Regards,
- David Turner
- The FreeType Project (www.freetype.org)
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- [Devel] Latest changes,
David Turner <=