0.22.0

This adds the capability to restart a codestream; restart allows the decoder/encoder to re-use memory that was allocated during previous usage of the codestream object.
This is useful for processing many similar codestream.

What's Changed

• Adds an option to restart a codestream by @aous72. See #194

Full Changelog: 0.21.5...0.22.0

0.21.5

This fixes a build failure with GCC 7.5.0 and 9.3.1.
Remember that 0.21.4 was an important update, because it added placeholder passes decoding.

What's Changed

• This fixes a build failure on GCC 9.3.1, and a warning. by @aous72 in #192

Full Changelog: 0.21.4...0.21.5

0.21.4

This is a small but an important update.
It is important because it adds proper decoding of placeholder passes; any HT compliant decoder should support them.
It also fixes a bug in avx512 encoder, as detailed in #188.

What's Changed

• This should fix the problem for the time being #188. by @aous72 in #190
• Placeholder passes by @aous72 in #191

Full Changelog: 0.21.3...0.21.4

0.21.3

This adds support for macOS multi-generation which can be invoked with -DCMAKE_OSX_ARCHITECTURES="arm64;x86_64".
This also adds the -fPIC flag to the library to generate position independent code.
There are not code changes, except to the enable the above.
This closes issue #184.

What's Changed

• Add support for macOS multi-generation by @aous72 in #185

Full Changelog: 0.21.2...0.21.3

0.21.2

This fixes a bug that was discovered recently; it might have been lurking in the background for a long time.
The bug results in coding all-zero codeblocks, producing files that are larger than they need to be.
The increase in size is not fixed and can change from one run to the other.
The bug is on the SIMD accelerated paths only.

This also include a PR from @kmilos -- thank you.

What's Changed

• CMake: polish export config by @kmilos in #182
• This fixes a bug that results in coding all-zero codeblocks by @aous72 in #183

Full Changelog: 0.21.1...0.21.2

0.21.1

This enables compilation for 32bit systems.
I never intended to support 32bit properly; so, this version allows compilation for 32bit systems with many warnings.
For Windows, you need the correct compiler.
For Linux you need the -m32 flag. Run export CXXFLAGS="-m32" and export CXXFLAGS="-m32" before running cmake.
To have TIFF support, you need the 32bit TIFF library.

What's Changed

• Update CMakeLists.txt for config file naming convention by @kmilos in #180
• Compilation for 32bit by @aous72 in #181

New Contributors

• @kmilos made their first contribution in #180

Full Changelog: 0.21.0...0.21.1

0.21.0

Many modification and changes -- see the log, but the important ones are:

1. Added lossy compression to data samples that are originally floating point numbers, and that are undergoing non-linear type3 transform, which stored with the help of the NLT marker segment.
2. Better support for components with differing properties; these involve using the QCC and COC marker segments. These are exposed using the API, but not in the applications.
3. OpenJPH can now be added using cmake's find_package().
4. For WASM, cmake configuration now generates one version only, with SIMD support; this support can be disabled through a command line option.

What's Changed

• Fixes a bug when a resolution has one 64bit line. by @aous72 in #162
• Makes all colour components involved in colour transform employ the s… by @aous72 in #163
• Allow OpenJPH to be exported by cmake by @palemieux in #166
• Supporting differing components in QCD by @aous72 in #169
• Adding NLT implementation to the lossy path by @aous72 in #170
• This adds support for COC. by @aous72 in #172
• Add support for find_package(), clean-up EMCC support and clean-up TIFF support by @palemieux in #173
• Adding Support for WASM without SIMD. by @aous72 in #174
• This is a better way of packet header parsing. by @aous72 in #176
• Support for lossy floating point with NLT segment marker. by @aous72 in #178

Full Changelog: 0.18.2...0.21.0

0.18.2

This is bug fix. This bug only occurred when the bit depth is large, as in more than around 28.

What's Changed

• This fixes a bug which wrongly sets implementation precision. by @aous72 in #161

Full Changelog: 0.18.1...0.18.2

0.18.1

This addresses the illegal instruction issue when -flto flag is used (link time optimization) as details in this thread
https://bugzilla.redhat.com/show_bug.cgi?id=2307795

What's Changed

• Addressing lto by @aous72 in #160

Full Changelog: 0.18.0...0.18.1

0.18.0

This release adds support for 32bit lossless compression of integer and floating-point data, where floating-point data can benefit from the non-linear (NLT) segment marker, which represent negative values in a more compression-friendly format.

This required the addition of 64 bit processing path for lossless compression, which is automatically employed if needed.

This version also addresses issue #157, which causes illegal instruction fault on machines that do not support AVX or more recent instruction sets, when the code in compiled with clang and in the Release mode only.

What's Changed

• Adding 32bit lossless by @aous72 in #158

Full Changelog: 0.17.0...0.18.0