Not that it matters, because JNI uses dlopen()/dlsym() to load symbols, but for bookkeeping purposes, move the JNI symbols introduced in libjpeg-turbo 1.3 into a new section.
Subtle point, but since libjpeg-turbo 1.3 now supports scaling factors > 1, the width and height specified in tjDecompress2() are not necessarily smaller than the JPEG width/height.
Fix I/O suspension. This little nugget of code was introduced in r30 as part of an early attempt to make buffered I/O work with the optimized Huffman codec. Ultimately, r32 reverted a lot of that mess and introduced much of the logic we now use, rendering this code unnecessary, but it was never reverted because it only causes problems when I/O suspension is used, and apparently no one has tried to do that with libjpeg-turbo until now.
Fix I/O suspension. This little nugget of code was introduced in r30 as part of an early attempt to make buffered I/O work with the optimized Huffman codec. Ultimately, r32 reverted a lot of that mess and introduced much of the logic we now use, rendering this code unnecessary, but it was never reverted because it only causes problems when I/O suspension is used, and apparently no one has tried to do that with libjpeg-turbo until now.
In order to avoid a functional regression with previous releases, the JAR file needs to be able to load either the 64-bit or 32-bit JNI library on Un*x systems.
Move the TurboJPEG DLLs back into the system directory on Windows platforms. For Windows, it doesn't really simplify the build system to install these libraries in c:\libjpeg-turbo*, and it introduces potential problems with loading the JNI library. Specifically, if a user linked their Java app against the 64-bit libjpeg-turbo SDK and then used a 32-bit JVM at run time, they would not be able to load the 32-bit turbojpeg.dll without manipulating java.library.path or the PATH environment (and vice versa for building against the 32-bit libjpeg-turbo SDK and using a 64-bit JVM at run time.)
Older versions of automake don't support multiple _JAVA primaries, but we don't actually have to specify a "dist" primary, since the Java sources are already included under EXTRA_DIST.
Further enhancements/fixes to the packaging system:
-- The Mac and Cygwin packages will now be created with the directory structure defined by the configure variables "prefix", "bindir", "libdir", etc., with the exception that the docs are always installed under /usr/share/doc/{package_name}-{version} on Cygwin and /Library/Documentation/{package_name} on Mac.
-- Fixed a duplicate filename warning when generating RPMs with the default prefix of /opt/libjpeg-turbo.
-- Moved the TurboJPEG libraries out of the system directory on Windows and Mac. It is no longer necessary to put them there, since we are not trying to be backward compatible with TurboJPEG/IPP anymore.
-- Fixed an issue whereby building the "installer" target on Windows would not build the Java JAR file, thus causing an error if the JAR had not been previously built.
-- Building the "install" target on Windows will now install libjpeg-turbo into c:\libjpeg-turbo[-gcc][64] (the same directories used by the installers.) This can be overridden by setting CMAKE_INSTALL_PREFIX.
-- The Java classes on all platforms will now look for the JNI library in the directory under which the build/packaging system installs it.
Overhaul Linux/Unix packaging system, primarily to avoid conflicts with vendor-supplied libjpeg-turbo packages (such as in Fedora and RHEL 6.) This also streamlines the packaging system somewhat, since it is no longer necessary to move the TurboJPEG libraries into the system library directory. Relocating those libraries was originally done to provide backward compatibility with TurboJPEG/IPP, but that package is long obsolete, and the software that formerly used it has been linking statically with libjpeg-turbo for quite some time.
If the default prefix (/opt/libjpeg-turbo) is used, then we now always install 32-bit libraries in /opt/libjpeg-turbo/lib32 and 64-bit libraries in /opt/libjpeg-turbo/lib64 instead of trying to conform to the Debian or Red Hat conventions. The RPM and DEB packages will now be created with the directory structure defined by the configure variables "prefix", "bindir", "libdir", etc., with the exception that the docs are always installed under /usr/share/doc/{package_name}-{version}.
DRC [Sat, 30 Mar 2013 20:54:15 +0000 (20:54 +0000)]
Remove unneeded headers (cdef.h isn't available on all platforms) and change u_int32_t to unsigned int, since u_int32_t isn't available on all platforms. This fixes build issues on Solaris.
DRC [Sat, 23 Mar 2013 05:35:37 +0000 (05:35 +0000)]
It's more correct to say "The Independent JPEG Group's software", since libjpeg-turbo emulates both the command-line and application programming interfaces.
DRC [Mon, 11 Feb 2013 03:38:53 +0000 (03:38 +0000)]
Trying to build ARM v7/v7s iOS binaries from the "official" libjpeg-turbo tarball (generated on RHEL4/autoconf 2.59/automake 1.9.2/libtool 1.5.6) using either Xcode 3.2.x or 4.5.x produces an error: "libtool: compile: unable to infer tagged configuration". Further study is warranted, but the most expedient way around it at the moment is to re-run autoconf on the OS X machine that is being used to generate the iOS binaries.
DRC [Wed, 6 Feb 2013 23:51:08 +0000 (23:51 +0000)]
Include a C version of md5cmp rather than depending on an external md5sum binary, since md5sum is not available on all platforms (specifically, it doesn't exist on FreeBSD, and it has to be installed via MacPorts on OS X.)
DRC [Mon, 21 Jan 2013 23:42:12 +0000 (23:42 +0000)]
Eliminated the awkward and confusing "TurboJPEG/OSS" designation, since there are no other active implementations of the TurboJPEG API anymore; don't refer to the libjpeg API library as "libjpeg-turbo" anymore, since that can be confusing; ARM v7s build instructions
DRC [Sat, 19 Jan 2013 06:43:27 +0000 (06:43 +0000)]
ImageIO.read() returns null if the input image type is not supported (which occurs when trying to read a PPM file), so output a friendly error instead of letting the next line throw a null pointer exception.
DRC [Sat, 19 Jan 2013 01:06:46 +0000 (01:06 +0000)]
Fix several issues with SRPM generation: (1) ensure that all relevant configure arguments get passed down to the configure command line in the generated spec file, (2) adjust the file manifest in the spec to accommodate the differing "age" version whenever the in-memory source/dest managers are used, and (3) fix an issue with the value of SO_MAJOR_VERSION passed down to the configure command line in the generated spec file (SO_MAJOR_VERSION has to remain pure, so we use a different variable to pass down the combined "current+age" value to libtool in Makefile.am.)
DRC [Sat, 19 Jan 2013 00:13:57 +0000 (00:13 +0000)]
Since Windows doesn't use lazy loading, clarify that a Windows application that uses jpeg_mem_dest() or jpeg_mem_src() must use the DLL containing those functions at run time. Other wordsmithing
DRC [Fri, 18 Jan 2013 06:02:10 +0000 (06:02 +0000)]
Subtle point, but the libjpeg v7+ API is not backward incompatible. That is, programs that were built against jpeg-6b can still build against jpeg-7+ with no issues. It's only the ABI that is backward incompatible, so the primary justification for implementing the emulation feature was to provide run-time (ABI) compatibility.
DRC [Fri, 18 Jan 2013 05:54:06 +0000 (05:54 +0000)]
Subtle point, but the libjpeg v7+ API is not backward incompatible. That is, programs that were built against jpeg-6b can still build against jpeg-7+ with no issues. It's only the ABI that is backward incompatible, so the primary justification for implementing the emulation feature was to provide run-time (ABI) compatibility.
DRC [Sun, 13 Jan 2013 12:15:58 +0000 (12:15 +0000)]
Fix the x86 build with NASM 0.98. Since NASM 0.98 is the default version on OS X, we want to at least allow people to build 32-bit code with it, even though it can't properly build 64-bit code.
DRC [Sun, 13 Jan 2013 12:12:53 +0000 (12:12 +0000)]
Fix the x86 build with NASM 0.98. Since NASM 0.98 is the default version on OS X, we want to at least allow people to build 32-bit code with it, even though it can't properly build 64-bit code.
DRC [Sun, 13 Jan 2013 11:25:20 +0000 (11:25 +0000)]
Consolidate the MD5 sums into one location and add a --without-turbojpeg switch to the Un*x build to allow building libjpeg-turbo without the TurboJPEG/OSS wrapper library. These modifications were supposed to lay the ground work for adding compile-time-selectable 12-bit JPEG support, but unfortunately there are deeper issues that prevent the implementation of that feature right now (namely, some of the modifications made to the C code to support the SIMD code are apparently not 12-bit-friendly.)
DRC [Sun, 13 Jan 2013 03:00:25 +0000 (03:00 +0000)]
Modify 'make test' so that it uses MD5 sums instead of reference images. This eliminates the need to check most of the test images into the repository, which keeps the source tarball to a reasonable size.
DRC [Sun, 13 Jan 2013 01:17:08 +0000 (01:17 +0000)]
install.txt contains no information relevant to libjpeg-turbo, and filelist.txt does not give any information that can't be gleaned by looking at the headers in the various source files.
DRC [Thu, 10 Jan 2013 11:47:39 +0000 (11:47 +0000)]
Wordsmithing; Remove mention of TurboJPEG/IPP-- it is no longer a relevant comparison, since the version of IPP on which TurboJPEG/IPP was based is now quite old, and TurboJPEG/IPP is no longer distributed or supported by The VirtualGL Project; Include information about mathematical incompatibilities with jpeg-8
projects / libjpeg-turbo / log