| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
| |
Add a version check to avoid depending on unstable GLib.
|
|
|
|
|
|
| |
gcc has optimizations for include guards that only work
if they are outermost in the the header.
https://bugzilla.gnome.org/show_bug.cgi?id=689810
|
|
|
|
| |
Coverity complained about this.
|
|
|
|
|
| |
I left out the updates to prepare the VS2010 install project for VS2012
in my last commit. :|
|
|
|
|
|
|
|
|
|
|
| |
-Prepare support for Visual Studio 2012, by adding the PlatformToolset tag
to all configs so that we can provide support for VS2012 with minimal
effort by simply copying and replacing the VS2010 items with the VS2012
counterparts with scripts.
-Improve consistency by using MultiByte character sets for all project
configs
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Parsing the orientation tag in big-endian file formats got broken in
commit 0179dfd as reported in https://bugs.launchpad.net/bugs/1077186
A short (16 bit) value like the orientation value occupies the two first
bytes of the 4-byte offset field, independently of endianness. The
commit changed the code from using de_get16() to de_get32 to read out
this value, which broke on big-endian (MM/Motorola) byte order JPEG file
formats. I believe little-endian files would still work as long as the
two last, unused bytes were zero.
From http://www.cipa.jp/english/hyoujunka/kikaku/pdf/DC-008-2010_E.pdf
"Value Offset
This tag records the offset from the start of the TIFF header to the
position where the value itself is recorded. In cases where the value
fits in 4 Bytes, the value itself is recorded. If the value is smaller
than 4 Bytes, the value is stored in the 4-Byte area starting from the
left, i.e., from the lower end of the byte offset area. For example, in
big endian format, if the type is SHORT and the value is 1, it is
recorded as 00010000.H."
Also avoid the confusing recycling of the "offset" variable, use a
better name for the second use and make it and its siblings
local to the block where they are used.
Signed-off-by: Tormod Volden <debian.tormod@gmail.com>
https://bugzilla.gnome.org/show_bug.cgi?id=688427
|
| |
|
| |
|
|
|
|
|
|
| |
AC_PATH_PROG is problematic for cross compliation.
Use AC_CHECK_TOOLS instead.
https://bugzilla.gnome.org/show_bug.cgi?id=671516
|
|
|
|
| |
This was pointed out in bug 686139.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=686605
|
|
|
|
|
|
|
|
| |
People on IRC were rightly pointing out that the fix is easy once you
know it. So why wouldn't gdk-pixbuf print it?
Also reinstate the warning. The pixdata module is always loaded so we
need to expect one existing module.
|
|
|
|
|
|
| |
Needed for the use of g_type_ensure().
https://bugzilla.gnome.org/show_bug.cgi?id=686844
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This call is necessary to ensure we actually link against libgobject;
otherwise it may be stripped if -Wl,--as-needed is in use.
The reason we need to link against libgobject is because it now has
a global constructor. If the dynamically loaded modules happen
to dlclose() libgobject, then reopen it again, we're in for trouble.
See https://bugzilla.gnome.org/show_bug.cgi?id=686822
Patch-suggested-by: Ryan Lortie <desrt@desrt.ca>
Signed-off-by: Colin Walters <walters@verbum.org>
https://bugzilla.gnome.org/show_bug.cgi?id=686822
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
The reason why the tiff loader was marked as not threadsafe is
that libtiff only has a global error handler; while we still have
to set an error handler to keep libtiff from spamming stderr,
we just give up on collecting any error messages from it, and
we also give up on the illusion that we can push/pop error handlers.
That doesn't work without a stack, so we just keep forcing our
do-nothing error handler on without ever taking it back.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|