Feature Changes:
- After the network manager was introduced to named to handle incoming traffic,
it was discovered that recursive performance had degraded compared to
previous BIND 9 versions. This has now been fixed by processing internal
tasks inside network manager worker threads, preventing resource contention
among two sets of threads.
- Zone dumping tasks are now run on separate asynchronous thread pools. This
change prevents zone dumping from blocking network I/O.
- inline-signing was incorrectly described as being inherited from the
options/view levels and was incorrectly accepted at those levels without
effect. This has been fixed; named.conf files with inline-signing at those
levels no longer load.
Full changelog:
https://bind9.readthedocs.io/en/v9_16/notes.html#notes-for-bind-9-16-17
This fixes a shift-reduce conflict in the parser that causes e.g.
"static const int" to be rejected as a syntax error.
This patch should have been handled by upstream, but there doesn't seem
to be an upstream.
Upstream decided to delete the distfile and let people download the
on-demand-created tarball instead. At least that seems to be generated
deterministically.
This is required to make 'export -f' work, which is enabled by default on
other OS I checked. While here sort the list of enabled options so that
it's easier to compare against the list of available options, as there are
still a number that we do not enable, and remove the bogus comment.
Bump PKGREVISION.
Classify ChIP/ATAC-Seq peaks based on features provided in a GFF
Peaks are provided in a BED file sorted by chromosome and position. The GFF
must be sorted by chromosome and position, with gene-level features separated
by ### tags and each gene organized into subfeatures such as transcripts and
exons. This is the default for common data sources.
CMake 3.20.4
* ci: use consistent sccache builds
* VS: Add special case for '-T version=14.29.16.10' under VS 16.10
* VS: Add flag table entries for '/external:W*' flags in VS 16.10
* gitlab-ci: Update Windows builds to MSVC 19.29-16.10 toolset
* Makefiles: Fix CMAKE_EXPORT_COMPILE_COMMANDS crash with custom compile rule
* presets: Fix buildPreset "jobs" field test case
* IRSL: Add Intel oneAPI redist location on Windows
* fileapi: Fix codemodel-v2 link command fragment relative paths
* FindBoost: Add check for json component header in Boost 1.75+
* Help: cmake_path: fix erroneous example for IS_PREFIX
* MSVC: C++20 final flag, C++23 support
* Clang/MSVC: C++20 final flag, C++23 support
* presets: Fix buildPreset "jobs"
* presets: Fix buildPreset "targets" not allowing a single string
1.81.1 June 11, 2021
* Added choice of codec (JP2 or J2K) when writing jp2k files.
* Fix use of hashmap with key based on dna.
1.81.0 June 6, 2021
* Fixed problems with tiff pdf wrapping photometry.
* Fixed scaling issues in prog/cleanpdf for printing.
* New progs: tiffpdftest, hashtest
* Fixed uninitialized data error in pixAddBorderGeneral() and
pixRemoveBorderGeneral()
* Rewrote Numa functions that discretize into bins. Have
binning by both sorting and histogram.
* Rewrote and simplified pixGetRankColorArray() and pixGetBinnedColor().
* Added tests to prog/rankbin_reg.c.
* Simplified fpixCopy() and dpixCopy(), and functions that use them.
* Check input for bilateral transforms.
* Add function for splitting a file evenly by lines.
* Check input for getFilenamesInDirectory()
* Many new fuzzers.
* Use size_t for all size inputs to ascii85 encoding/decoding
* New regression tests: encoding_reg.c, binmorph6_reg.c,
flipdetect_reg.c
* Reworked concatpdf for generality, using the Poppler package.
* Removed dwa flipdetection from the library. All the dwa code
is now in flipdetectdwa.c.notused. Likewise prog/flipselgen.c
is retained for completeness, renamed flipselgen.c.notused,
and is not compiled.
* Added hashmap utility (hashmap.c, hashmap.h).
* Removed functions using dnahash on strings, pts and doubles.
* Improved the speed of hashing for strings and doubles.
* Added function for tiling images in pairs for comparison.
* Added null terminations to serialized strings written to memory,
preventing buffer overrun by strlen() [fixed by Stefan Weil].
* No longer use "NoInit" versions of pix creation in leptonica.
because they risk reading uninitialized data. These functions
remain in the library because they are in use in applications.
* Add two composite binarization functions, from prog/binarize_set.
* Giulio Lunati fixed pnm reading to work with stdin input.
* Removed several of the boxa compare and modify functions.
* Implemented reading jpeg200 data encded in j2k "codestream" format.
Can now read data in both jp2 and j2k.
