Tony Cook [Tue, 26 Apr 2005 10:45:21 +0000 (10:45 +0000)]
- Imager::Font::BBox objects now have right_bearing() and display_width()
methods. Hopefully this and the advance_width() method fills out
the Imager bounding box interface to a useful state.
Implemented for all four font drivers.
- Win32 font bounding_box() method now supports the advance width
and right bearing values.
Tony Cook [Tue, 26 Apr 2005 08:55:35 +0000 (08:55 +0000)]
- Imager::Font::BBox objects now have right_bearing() and display_width()
methods. Hopefully this and the advance_width() method fills out
the Imager bounding box interface to a useful state.
Implemented for FT2, FT1.x, T1Lib so far.
Tony Cook [Sun, 24 Apr 2005 11:28:01 +0000 (11:28 +0000)]
- Imager::Font::BBox objects now have right_bearing() and display_width()
methods. Hopefully this and the advance_width() method fills out
the Imager bounding box interface to a useful state.
Implemented for FT2 so far.
Tony Cook [Thu, 21 Apr 2005 14:58:55 +0000 (14:58 +0000)]
- the FT 1.x was comparing versus an uninitialized variable when
checking the end of string when calculating the right side bearing
used to adjust pos_width for glyphs that overlap the right side of the
advance width.
Tony Cook [Wed, 20 Apr 2005 15:21:29 +0000 (15:21 +0000)]
- Imager::Font::BBox advance_width() method was falling back to
the wrong value when the font driver's low level function
didn't return the advance width.
Tony Cook [Wed, 20 Apr 2005 13:38:40 +0000 (13:38 +0000)]
- the segments parameter supplied to the fountain filter and the
fountain fill contructor now accepts color names or other
descriptions as other functions do.
Tony Cook [Fri, 15 Apr 2005 10:48:27 +0000 (10:48 +0000)]
- newer versions of GIMP add a line to gradient files before the
segment count giving a descriptive name of the gradient.
Imager::Fountain can now read and write such gradient files. The
interface is a bit indirect, but I'd like to preserve
Imager::Fountain as a blessed array ref for now.
Tony Cook [Wed, 16 Mar 2005 13:48:51 +0000 (13:48 +0000)]
- added README for samples directory, describes interleave.pl and
anaglyph.pl to start.
- the XS for the internal i_glin() function wasn't copying the pixel
values across to the returned color objects.
http://rt.cpan.org/NoAuth/Bug.html?id=11860
Tony Cook [Fri, 11 Mar 2005 11:57:34 +0000 (11:57 +0000)]
- the convert, crop, rotate, copy, matrix_transform, to_paletted, to_rgb8,
scaleX and scaleY methods now warn when called in void context.
http://rt.cpan.org/NoAuth/Bug.html?id=9672
Tony Cook [Tue, 8 Mar 2005 08:02:59 +0000 (08:02 +0000)]
- calling the read() method for a format not included in the Imager build,
for example, JPEG with no libjpeg installed, would crash with an
undefined function error (modified by DynaLoaders dependence on
AutoLoader.
http://rt.cpan.org/NoAuth/Bug.html?id=9618
- some test scripts have been modified to use Test::More, which is now
included under the t directory. Eventually all will be modified to use
Test::More and the duplicates in t/testtools.pl will be removed
Tony Cook [Mon, 28 Feb 2005 06:56:27 +0000 (06:56 +0000)]
- reading an image with an idstring of 128 or more bytes would result
in an allocation error, if the platform char type was signed
- tests now check that tga tags are set
Tony Cook [Thu, 10 Feb 2005 13:12:56 +0000 (13:12 +0000)]
- call read() instead of open() in the example code in Imager.pm,
and mention that open() is an alias for read().
http://rt.cpan.org/NoAuth/Bug.html?id=11431
- added reference list of crop() parameters.
http://rt.cpan.org/NoAuth/Bug.html?id=11430
Tony Cook [Mon, 31 Jan 2005 14:08:57 +0000 (14:08 +0000)]
- the right-side bearing used to adjust the final width wasn't being
calculated correctly, this would cause the bounding_box() function for
freetype 2.x to return a larger than expected text width.
http://rt.cpan.org/NoAuth/Bug.html?id=11291
Tony Cook [Mon, 24 Jan 2005 10:19:50 +0000 (10:19 +0000)]
- scanning for required include files is now down by directory rather
than by file, using -e to check for the existence for a file rather than
doing an eq against every filename in the directory.
- previously we only checked a candidate include directory for freetype.h
to consider it a valid FT1.x include directory. We now check if it
contains fterrors.h too, if it does, then it's from a FT2 installation
and we ignore it.
Tony Cook [Thu, 23 Dec 2004 23:01:27 +0000 (23:01 +0000)]
- we were undefing the wrong macro (used as an include guard) when
building the error code translation function for freetype 2. This
meant most errors came out as numbers.
Tony Cook [Thu, 23 Dec 2004 01:05:44 +0000 (01:05 +0000)]
- Makefile.PL now checks the directories specified by $Config{locincpth}
and $Config{loclibpth} for includes and libraries respectively.
Resolves: https://rt.cpan.org/Ticket/Display.html?id=2942