v56.0.0
Deprecations
* The license_file option is now marked as deprecated. Use license_files instead.
Breaking Changes
* If neither license_file nor license_files is specified, the sdist option will now auto-include files that match the following patterns: LICEN[CS]E*, COPYING*, NOTICE*, AUTHORS*. This matches the behavior of bdist_wheel.
Changes
* The license_file and license_files options now support glob patterns.
* Implemented VendorImporter.find_spec() method to get rid of ImportWarning that Python 3.10 emits when only the old-style importer hooks are present
Documentation changes
* Added documentation for the license_files option.
v55.0.0
Breaking Changes
* Remove the deprecated bdist_wininst command. Binary packages should be built as wheels instead.
Version 4.3 (19 Jan 2020)
A complete list of bugs fixed in this version is available here:
https://sv.gnu.org/bugs/index.php?group=make&report_id=111&fix_release_id=108&set=custom
* WARNING: Backward-incompatibility!
Number signs (#) appearing inside a macro reference or function invocation
no longer introduce comments and should not be escaped with backslashes:
thus a call such as:
foo := $(shell echo '#')
is legal. Previously the number sign needed to be escaped, for example:
foo := $(shell echo '\#')
Now this latter will resolve to "\#". If you want to write makefiles
portable to both versions, assign the number sign to a variable:
H := \#
foo := $(shell echo '$H')
This was claimed to be fixed in 3.81, but wasn't, for some reason.
To detect this change search for 'nocomment' in the .FEATURES variable.
* WARNING: Backward-incompatibility!
Previously appending using '+=' to an empty variable would result in a value
starting with a space. Now the initial space is only added if the variable
already contains some value. Similarly, appending an empty string does not
add a trailing space.
* NOTE: Deprecated behavior.
Contrary to the documentation, suffix rules with prerequisites are being
treated BOTH as simple targets AND as pattern rules. Further, the
prerequisites are ignored by the pattern rules. POSIX specifies that in
order to be a suffix rule there can be no prerequisites defined. In this
release if POSIX mode is enabled then rules with prerequisites cannot be
suffix rules. If POSIX mode is not enabled then the previous behavior is
preserved (a pattern rule with no extra prerequisites is created) AND a
warning about this behavior is generated:
warning: ignoring prerequisites on suffix rule definition
The POSIX behavior will be adopted as the only behavior in a future release
of GNU make so please resolve any warnings.
* New feature: Grouped explicit targets
Pattern rules have always had the ability to generate multiple targets with
a single invocation of the recipe. It's now possible to declare that an
explicit rule generates multiple targets with a single invocation. To use
this, replace the ":" token with "&:" in the rule. To detect this feature
search for 'grouped-target' in the .FEATURES special variable.
Implementation contributed by Kaz Kylheku <kaz@kylheku.com>
* New feature: .EXTRA_PREREQS variable
Words in this variable are considered prerequisites of targets but they are
not added to any of the automatic variable values when expanding the
recipe. This variable can either be global (applies to all targets) or
a target-specific variable. To detect this feature search for 'extra-prereqs'
in the .FEATURES special variable.
Implementation contributed by Christof Warlich <cwarlich@gmx.de>
* Makefiles can now specify the '-j' option in their MAKEFLAGS variable and
this will cause make to enable that parallelism mode.
* GNU make will now use posix_spawn() on systems where it is available.
If you prefer to use fork/exec even on systems where posix_spawn() is
present, you can use the --disable-posix-spawn option to configure.
Implementation contributed by Aron Barath <baratharon@caesar.elte.hu>
* Error messages printed when invoking non-existent commands have been cleaned
up and made consistent.
* The previous limit of 63 jobs under -jN on MS-Windows is now
increased to 4095. That limit includes the subprocess started by
the $(shell) function.
* A new option --no-silent has been added, that cancels the effect of the
-s/--silent/--quiet flag.
* A new option -E has been added as a short alias for --eval.
* All wildcard expansion within GNU make, including $(wildcard ...), will sort
the results. See https://savannah.gnu.org/bugs/index.php?52076
* Interoperate with newer GNU libc and musl C runtime libraries.
* Performance improvements provided by Paolo Bonzini <pbonzini@redhat.com>
GNU make Developer News
* Import the GNU standard bootstrap script to replace the hand-rolled
"make update" method for building code from a GNU make Git repository.
* Rework the source distribution to move source files into the src/*
subdirectory. This aligns with modern best practices in GNU.
* Replace local portability code with Gnulib content. Unfortunately due to a
problem with Gnulib support for getloadavg, this forces a requirement on
Automake 1.16 or above in order to build from Git. See README.git.
Release Notes for cloc version 1.90:
New Languages and File Types:
o BizTalk Orchestration
o BizTalk Pipeline
o Cake Build Script
o C# Designer
o CSV
o Delphi Form
o Gleam
o Godot Resource
o Godot Scene
o Juniper Junos
o ReScript
o Ring
o Visual Studio Solution
o Web Services Description
o Zig
Updates:
o Dockerfile: add .[dD]ockerfile as an extension; treat
files named Dockerfile.ext as Dockerfile unless ext
maps to a recognized language other than cmake or m4.
o Identify autogenerated C# files.
o If getting input from a list file or using a language
definition file, also check for (and use, if it exists)
an options.txt file in the same directory.
o Use --csv-delimiter for --ignored, --categorized output.
o Support diff summation in csv format.
o Update cloc's Unix Makefile to support macOS
o Rename Junos to Juniper Junos
o Stratify Visual Basic to for Applications, .NET, Script
o Use tar on Windows 10 instead of zip with --git --diff
o Track upper/lowercase file names on Windows to permit
consistent diffs.
o Support nested Elm comments.
o Update to Regexp::Common 2017060201.
o Minimum Perl version is now 5.10.0
o New options --skip-leading, --summary-cutoff
o Refine comment definitions for ASP.NET, Razor
o Handle Julia docstrings.
o Adjust timeout and diff_timeout values to avoid alarm
overflows.
o Add extensions .btproj, .msbuild for MSBuild script
Bug Fixes:
o Add missing Raku_or_Prolog() subroutine.
o Handle UTF encoded list files (file given to --list-file).
o Improved support for options parsing from config file.
o Catch unidentified files via --diff-list-file.
o Don't insert extra newlines when stripping with --unicode.
o Strip literal '\x' from JSON output strings.
o Improve logic for detecting Unicode files less than
90 bytes in size.
o Fix divide by zero classifying nonreadable .b files.
o Escape backslash for 'report_file' entry in JSON, YAML on
Windows.
Now it should be more obvious when a package needs it as a dependency,
as it will fail loudly if it isn't declared as a tool.
While here, some duplicate dependencies on itstool were removed from the
MATE packages
## What's new in LuaRocks 3.7.0
* Improved connectivity resiliency
* LuaRocks can now use mirrors for downloading rocks even if downloading
the manifest from the main server succeeds.
In previous versions, LuaRocks would check whether to use a mirror in the first
download operation, when it fetches the manifest. Once the server
(luarocks.org or one of its default mirrors) was chosen, it would stick with
it for the rest of the command.
The resulting behavior was that if the manifest fails to load, it switches to
a mirror and continues from there. But if the manifest fetches ok and the then
actual rock download fails, it would give up, instead of trying that in a
mirror as well.
Now, it retries every download on a mirror whenever the base URL matches one
configured in cfg.rocks_servers. The original behavior was satisfactory if
there was complete downtime in the main server, but this new behavior should
make the CLI much more resilient with regard to any intermittent failures
happening on the main server.
* On Unix, it now respects environment variables $XDG_CACHE_HOME and $XDG_CONFIG_HOME
* This means the user's configuration typically resides in ~/.config/luarocks/
as per the XDG standard
* The legacy path ~/.luarocks/ continues to be tested first, for backwards
compatibility
* Fixes check for the default Lua version set in the user's home configuration
* Fixes an issue on Windows where it would incorrectly revoke permissions
from the current user when installing
## What's new in LuaRocks 3.6.0
* Adds a double-check step to verify that all files from a rock are installed
* Improve resilience of the manifest reader to deal with manifests
written with older versions of LuaRocks lower than 3.0
* `luarocks pack` now checks that the directory inside the archive being packed
as a `.src.rock` actually exists, refusing to pack an invalid rock from
a badly configured rockspec.
* Fixes behavior of `luarocks pack` when the `url` entry of a rockspec
points to a bare file.
* Remove an entry from the manifest if the rock itself is already missing
* The `configure` script now checks that the version of `lua.h`
found matches that of the Lua interpreter detected or configured
* Fixes the renaming of scripts when multiple versions are installed
* Fixes availability check for `svn` for rockspecs using Subversion
* Fixes for running with an empty PATH environment variable
* Portability improvements:
* Windows: vcvarsall.bat output is now properly redirected to NUL
meaning that the output of `luarocks path` can be used in scripts
* Fixes autodetection for Cygwin
* Handles macOS versions greater than 10.10
* Adds platform specific configurations for NetBSD
* Respects CC/CFLAGS/LDFLAGS on FreeBSD
* Luacheck now runs on the LuaRocks CI
* Distributed binaries are built using Lua 5.3
0.19 2021-04-23
- We now prefer the mesh() and zip() functions from List::SomeUtils over the
new, incompatible implementations in List::Util 1.56. Fixes GH #12.
reported by Greg Oschwald.
Upstream changes:
0.080 2021-04-16 T. R. Wyant
All uses of the postderef argument to new() now warn.
0.079 2021-03-26 T. R. Wyant
Get prerequisites up to snuff, and add xt/author/prereq.t to ensure
they stay that way.
Add rt.cpan.org back to bug reporting methods. Long live RT!
0.078 2021-01-28 T. R. Wyant
Allow CPAN to index Script_Run, Atomic_Script_Run, since they made
it into a production release.
Allow {,3} and { 0 , 3 } as quantifiers, requiring at least Perl
5.33.6. Previously these parsed as literals. This parse will be
retracted if it does not make it into 5.34.0.
0.077 2021-01-14 T. R. Wyant
Add Travis CI testing.
Use GitHub as bug tracker. R.I.P. rt.cpan.org.
0.076 2020-11-28 T. R. Wyant
Correct (I hope) detection of \K in nested assertions.
Variable-length look-behind is version 5.029009.
Look-behinds quantified longer than 255 characters are an error, and
are made into unknown tokens or structures. I ended up refactoring
the PPIx::Regexp::Token::GroupType class initialization for the
latter two changes.
0.075 2020-10-08 T. R. Wyant
Warn on first use of attribute 'postderef'.
Upstream changes:
0.109 2021-03-16 19:55:45-04:00 America/New_York
- eliminate warnings-count failure by requiring an ExtUtils::MakeMaker
from late 2013 or later; without this, very old EUMM could pass -w
to the tests, enabling more warnings than we wanted (thanks, Matthew
Horsfall and Graham Knop!)
0.108 2021-03-16 09:54:51-04:00 America/New_York
- provide diagnostics in tests when more warnings arrive than are
expected
0.107 2021-03-14 16:15:57-04:00 America/New_York
- Term::ReadKey has been dropped; caused too many problems
- minimum version is now v5.10.1, not v5.10.0
0.106 2021-03-12 21:29:54-05:00 America/New_York
- improved formatting of switches
- when available, use Term::ReadKey to get terminal width
- when an option name is defined twice, warn about it
THIS WILL BECOME FATAL IN A FUTURE VERSION
Upstream changes:
1.08 Fri Oct 09 17:45:09
- Added Data::Munge to SEE ALSO section.
- Added documentation for find_home.
- Skipping find_home test when Nigel's overly-restrictive environment detected.
Upstream changes:
0.12 2021-03-20 NEILB
- Fix for RT#128243, where touching a file in close succession wouldn't
register the second one for make & other purposes, on systems that
have finer granularity than seconds. Thanks to Slaven Rezic for
the suggested fix.
- Improved opening paragraphs of the DESCRIPTION, including a suggestion
to always require 0.12 or higher.
0.11_03 2021-03-18 NEILB
- Time::HiRes doesn't provide utime() on Windows, so now we try and
load Time::HiRes in a BEGIN block, and only enforce the min version
if it loaded ok.
0.11_02 2021-03-18 NEILB
- Looks like I should have specified a min version of
Time::HiRes. Was getting some fails from CPAN Testers,
about utime() not being available.
0.11_01 2021-03-17 NEILB
- Try Slaven's suggested fix for RT#128243
Upstream changes:
version 2.08 at 2021-01-11 20:08:00 +0000
-----------------------------------------
Change: 2e710bc17736e5d2c43875c2076db7c1a6e4fe47
Author: Chris 'BinGOs' Williams <chris@bingosnet.co.uk>
Date : 2021-01-11 20:08:00 +0000
This is 2.08
Change: 1eaa7976957abbc533869176a390638e93850d12
Author: Graham Knop <haarg@haarg.org>
Date : 2021-01-11 08:47:00 +0000
fix darwin DynaLoader patch for older perls
The patch that was taken from blead uses a <<$^OS-eq-darwin>>
preprocessing token, which is handled by the DynaLoader_pm.PL script.
Older versions of DynaLoader_pm.PL don't include this preprocessing,
but the patch still applies.
In our case, the preprocessing is a check for darwin. But the patch
is already only applied on darwin, so it is redundant. We can just
remove the preprocessing token, and it should work across all
versions.
Upstream changes:
Changes in Devel::NYTProf 6.08 - DATE TO COME
B<There should be no observable differences in the performance of Devel-NYTProf
from that of the previous release.> This release is maintenance-focused and
changes mainly concern improvements in test coverage, elimination of
build-time warnings and correction of small errors detected by end-users.
=head3 Substantive
Corrected 1 error in F<Makefile.PL>
(L<https://github.com/timbunce/devel-nytprof/pull/158>; thanks to Kent Fredric).
Corrected other errors in F<NYTProf.xs> and F<ppport.h> at suggestion of Reini Urban
in earlier pull request.
Two subroutines in F<Devel::NYTProf::Data> -- C<packages_at_depth_subinfo()>
and C<package_fids()> -- which were not used anywhere in the codebase or test
suite were removed and placed in F<devstuff/superseded.pm>. One function --
C<get_str_id()> defined in F<NYTProf.xs> but never used in the codebase was
similarly removed and placed in F<devstuff/superseded.xs>. Anyone who was
using these functions may file a pull request to have them restored to the
codebase.
In C<Devel::NYTProf::Data::strip_prefix_from_paths()>, a branch concerned with
the possibility that an argument was a hash was not exercised anywhere in the
codebase or the test suite and was simply commented out. Anyone who was using
this functionality may file a pull request to have it restored to the
codebase.
Eliminated all build-time warnings being generated during F<make> on testing
platforms Linux, FreeBSD, OpenBSD, Windows.
=head3 Test Suite
Improved coverage provided by test suite to each of the following files:
lib/Devel/NYTProf/Data.pm
lib/Devel/NYTProf/FileInfo.pm
lib/Devel/NYTProf/Reader.pm
lib/Devel/NYTProf/SubInfo.pm
lib/Devel/NYTProf/Util.pm
This was accomplished mainly by adding 4 new test files:
t/11-reader.t
t/12-data.t
t/13-fileinfo.t
t/14-subinfo.t
... and their corresponding data files. In addition, there were substantial
additions to F<t/30-util.t>.
Coverage was evaluated with C<trace_level> unset and C<trace_level> 5.
=head3 Housekeeping
Updated Travis and AppVeyor configuration files.
Updated F<MANIFEST> and F<MANIFEST.SKIP> so that F<make manifest> works as intended.
Moved the following 6 test files from F<t/> to F<xt/>:
t/68-hashline.t
t/71-moose.t
t/72-autodie.t
t/90-pod.t
t/91-pod_coverage.t
t/92-file_port.t
Reason: They either (a) had been previously described as insufficiently
developed and were being skipped; or (b) are of concern only to NYTProf
developers. Hence, a failure in any of them should not preclude installation
and use of Devel::NYTProf.
Added a F<make test_short> target to F<Makefile.PL> and environmental variable
C<NYTPROF_TEST_SHORT>. These will enable NYTProf developers to cut testing
time without significant loss in accuracy
(L<https://github.com/timbunce/devel-nytprof/pull/155>).
Corrected one date error in 6.07 release
(L<https://github.com/timbunce/devel-nytprof/issues/152>; thanks to @shawnlaffan).
Applied several commits either cherry-picked from, or initiated by Reini Urban
in an older pull request.
=head2 Changes in Devel::NYTProf 6.07 - 6th April 2021
Focus is to enable test suite to pass with perl-5.33.3 and higher. (Changes
in blead required changes in expectations set in test data files.)
Correction to MANIFEST
thanks to mattlaw. #142
POD improvements
thanks to tomhukins. #129
Makefile.PL modernizations
thanks to karenetheridge. #126
.gitignore corrections
thanks to jkeenan. #146
Adapt tests to work with 5.33 and higher
thanks to jkeenan. #147
Abseil LTS 20210324.1
What's New:
The cleanup library has been released. This library contains the control-flow-construct-like type absl::Cleanup which is used for executing a callback on scope exit.
The numeric library now includes bits.h, a polyfill header containing implementations of C++20's bitwise math functions.
Abseil now installs pkg-config files to make it easier to use Abseil with some other build systems.
Abseil now respects the default CMake installation paths. Standard CMake variables like CMAKE_INSTALL_PREFIX can be used to change the installation path.
Breaking Changes:
The empty absl::container target has been removed from the CMake build. This target had no effect and references to this target in user code can safely be removed.
CMake 3.20.2
* ci: pay attention to the machine load when running tests
* ci: limit builds by machine load
* Help: Add 3.20 release note for error on unknown arguments
* gitlab-ci: equally delay all jobs on integration branches
* Autogen: Restore mocs_compilation in OBJECT libraries
* Help: Fix typos in cmake-compile-features(7)
* Help: Do not recommend WCDH in cmake-compile-features(7)
* Intel: Update Classic compiler version detection for 2021
* IntelLLVM: Add special case for ifx 2021.1 version extraction
* Help: Behavior of file(TO_NATIVE_PATH) depends on the host platform
* Help: Document special cases for if(IS_ABSOLUTE)
* Add missing 'not' in error messages
* KWIML 2021-04-21 (49d91529)
* autogen: fix race in depfile parsing
* Ninja Multi-Config: Correctly generate POST_BUILD custom targets
* BinUtils: Restore toolchain prefix detection from compiler name 'c++'
* OpenBSD: Fix system feature definitions
0.19.0
- Fix AttributeError lineno
- Python 3.9 ast changes
- create_generate with goal_resource param
- Fix relative import offset calculation
- Fix missinge lineno attribute for AssignedName ast node
- Added _NamedExpr into `patchedast.py`
- Add support for the walrus operator.
- fix test case name for `test_ann_assign_node_without_target`
- Returned _AnnAssign and checked for support assignment without value
- fixed version restriction in tests for NamedExpr
- Removed AnnAssign, added NeamedExpr, testa are made
- Added _AnnAsign into `patchedast.py`
- Extract augmented assignment
- Fix handling of dict rename in Python 2.x
- Improve handling of generalized dict unpacking during dict rename
- Add expected failure test for comprehension variable scopes
- Implement basic scoping and rename for set and dict comprehension
- Visit subexpressions of comprehensions to collect names for scopes
- Implement rename of inline assignment expression
- Implement basic scoping and renaming of list and generator
comprehension loop variables
- Implement f-string extract refactoring
- Refactor consume_joined_string and also fix missing
ast.JoinedStr/FormattedValue in older python
- Fix some f-string corner cases
- Implement PEP-448 generalized dict-unpacking
Fix a typo
Upstream changes:
rlang 0.4.10
New hash() function to generate 128-bit hashes for arbitrary R objects using the xxHash library. The implementation is modeled after xxhashlite, created by @coolbutuseless.
New check_installed() function. Unlike is_installed(), it asks the user whether to install missing packages. If the user accepts, the packages are installed with pak::pkg_install() if available, or utils::install.packages() otherwise. If the session is non interactive or if the user chooses not to install the packages, the current evaluation is aborted (#1075).
rlang is now licensed as MIT (#1063).
Fixed an issue causing extra empty lines in inform() messages with .frequency (#1076, @schloerke).
expr_deparse() now correctly wraps code using :: and ::: (#1072, @krlmlr).
Version 2.9.1
Bugfixes
* The internal locale-data loading functions now validate the name of the locale file to be loaded and only
allow files within Babel's data directory.
- Minor change in taint extraction: Check for non-whitespace in library
path after m{ (.+) }x, remove 's' in regex to avoid issues with paths
including embedded newlines. Up the minor number: if anyone really
does depend on locating all-whitespace paths or ones with embedded
newlines warn me.
1.9.0
=====
New Features
------------
- Add python type hints for python 3.
Bug Fixes
---------
- Issue 144 - When performing a read we handled an empty byte return as an indication to close the file pointer. This is not true when you actually ask for a zero byte read so we should allow it.
- Fix: When you have a non url-safe character in your URL it is quoted by requests however requests-mock just treated it as a normal string.
Upstream release notes:
- Security: Fixed command injection vulnerability in HgDriver/HgDownloader and
hardened other VCS drivers and downloaders (GHSA-h5h8-pc6h-jvvx /
CVE-2021-29472)
- Fixed install step at the end of the init command to take new dependencies
into account correctly
- Fixed update --lock listing updates which were not really happening (#9812)
- Fixed support for --no-dev combined with --locked in outdated and show
commands (#9788)
What's New in Pylint 2.8.2?
Keep __pkginfo__.numversion a tuple to avoid breaking pylint-django.
scm_setuptools has been added to the packaging.
Pylint's tags are now the standard form vX.Y.Z and not pylint-X.Y.Z anymore.
1.012001 2021-01-10
[ Test Suite ]
- Extra test cases to improve coverage. (100% on Coveralls; 90% on
Codecov.)
- Hide warnings in Kavorka integration tests.
[ Packaging ]
- Move issue tracker from RT to Github Issues.
- Stop hiding Type::Parser::Token, Type::Parser::TokenStream, and
Type::Parser::AstBuilder from the CPAN indexer.
[ Other ]
- Much code tidying using perltidy and manually.
- When generic validation objects (blessed objects with a `check` method)
are converted to native Type::Tiny type constraints, no longer require
them to provide a `get_message` method. This allows Type::Tiny to adopt
Data::Constraint type constraints.
1.012000 2020-10-28
[ Documentation ]
- Update NEWS file.
1.011_011 2020-10-16
[ Test Suite ]
- Bugs in old versions of Return::Type prevent integration tests from
passing on Perl 5.8.x; those tests now require Return::Type 0.007 or
above.
- More tests for Type::Coercion::FromMoose.
[ Packaging ]
- If the EXTENDED_TESTING environment variable is true, Perl 5.8.9 or
above is being used, and either Type::Tiny's version contains '_' or
testing is running on Travis CI, then Makefile.PL will add extra testing
dependencies.
[ Other ]
- Remove unnecessary BEGIN block in Eval::TypeTiny.
1.011_010 2020-10-16
- Simple useful coercions from Str for Type::Tiny::Enum (and
Types::Standard::Enum).
1.011_009 2020-10-09
[ Documentation ]
- Add _ForeignTypeConstraint to AllTypes.pod.
[ Test Suite ]
- Add tests for _ForeignTypeConstraint.
- Improve test coverage by adding tests for various esoteric parts of the
distribution and edge cases.
[ Other ]
- More efficient use of Type::Tiny::XS by Types::TypeTiny.
- Refactoring of Types::TypeTiny.
- Where various parts of Type::Tiny internally use type constraints to
check values, prefer is_* and assert_* functions over method calls.
1.011_008 2020-10-07
[ Documentation ]
- Some minor tidyups and updates to Type::Tiny::Manual.
[ Test Suite ]
- Better tests for the placeholder type constraints created by
Type::Library -declare flag.
- Test integration with Type::Nano as an example of a generic
non-Type::Tiny type constraint class.
- Test that $type->() works with subclasses that override the
assert_return method.
- Write tests for some internal undocumented methods.
[ Other ]
- Added: Type::Utils::assert() function.
1.011_007 2020-10-06
[ Bug Fixes ]
- ArgsObject inlining was closing over a coderef in a way that didn't work
on archaic versions of Perl. Resolve by calling the coderef outside the
closure.
1.011_006 2020-10-02
[ Documentation ]
- Include JSONCapable type example.
[ Test Suite ]
- Improved test coverage for Type::Library.
- Improved test coverage for Type::Params.
- Improved test coverage for Type::Registry.
- Improved test coverage for Type::Tiny::Union.
- Improved tests for Type::Utils::is().
- Various tests for garbage collection using Devel::Refcount.
[ Other ]
- Added: Type::Params now exports an ArgsObject type constraint.
- Test::TypeTiny's should_pass and should_fail exercise type constraints
in even more ways in EXTENDED_TESTING mode.
1.011_005 2020-09-30
- Added: Type::Library import now supports -extends and -utils flags.
- Type::Library -base import flag now marks the caller package as loaded
in %INC.
1.011_004 2020-09-30
[ Bug Fixes ]
- Make predeclared type constraints work better for Zydeco.
[ Documentation ]
- Document that ArrayLike and HashLike are now parameterizable.
[ Test Suite ]
- Make test suite work better with App::Yath.
1.011_003 2020-09-25
[ Bug Fixes ]
- Old versions of Data::Dumper would sometimes die when dumping certain
overloaded objects. Type::Tiny::_dd() now catches this in an eval {}.
- Types::Standard would sometimes complain about very old versions of
Scalar::Util.
[ Other ]
- ArrayLike and HashLike are now parameterizable, though the
implementation for the parameterized version is in pure Perl, not XS.
- Type::Tiny::Enum better caches regexps.
- Updated: ArrayLike, HashLike, CodeLike, and StringLike type constraints
can use XS if Type::Tiny::XS 0.022 is installed.
1.011_002 2020-09-22
- Added: Type::Utils now exports an `is` function but it needs to be
requested explicitly.
1.011_001 2020-09-21
[ Documentation ]
- Update the NEWS file.
[ Other ]
- Added: Add new list processing functions to Type::Tiny.
1.011_000 2020-09-15
[ Documentation ]
- Describe deficiencies of is_* functions for parameterized types.
Fixes RT#132918.
<https://rt.cpan.org/Public/Bug/Display.html?id=132918>
[ Other ]
- Type::Tiny::Enum now generates faster regexps to validate enums. (Code
stolen from Regexp::Trie.)
1.033 2021-02-11T00:21:54Z
* v1.032 had some regressions because I released 1.031 from an
unmerged branch. https://github.com/briandfoy/test-output/issues/8
This was noticed by Tina Müller in https://github.com/os-autoinst/os-autoinst/pull/1622
* I didn't realize that I'd released an experiment in 2017, but
no own complained until it was reverted (by releasing from master).
The experiment worked I guess? Sub::Exporter is gone, so the
dependency tree for this is greatly reduced, and now it's gone
again.
1.032 2021-01-28T22:15:07Z
* Fix cut-n-paste error in docs (Github #6, hexcoder)
1.40 2021-02-23
- Migrate the repository and the bug tracker to GitHub
1.39 2021-02-22
- Makefile.PL
- some more fix for Homebrew on macOS [rt.cpan.org #133846]
1.38 2021-02-22
- Makefile.PL
- run 'brew --prefix ncurses' internally to detect ncurses installed
by Homebrew on macOS [rt.cpan.org #133846]
- utilize ExtUtils::Liblist::ext() [rt.cpan.org #134408]
On macOS Big Sur or later 'perl Makefile.PL' may warn as
"Could not find neither libtermcap, libtinfo, libncurses, or libcurses."
You should be able to ignore this warning if you have Xcode installed. See
https://github.com/Perl-Toolchain-Gang/ExtUtils-MakeMaker/issues/381
for details. To suppress this message, use ExtUtils::MakeMaker 7.58 or later.
- can specify multiple directories on --prefix, --libdir, and
--includedir options by separating them with colons
- INSTALL
- update for changes above and so on.
1.37 2020-12-27
- readline-8.1 support
new function
rl_activate_mark
rl_deactivate_mark
rl_keep_mark_active
rl_mark_active_p
use native operate_and_get_next on GNU ReadLine 8.1 and later.
- Gnu.xs: warns if attempted completion function returns undef in pos 0.
[rt.cpan.org #132384]
- ppport.h: update from version 3.31 to version 3.52.
- fixes to be more CPAN Testers friendly
- Makefile.PL exits 0 when $TERM is not set properly. [rt.cpan.org #128806]
- update URLs of the GNU Readline Library manual website. [rt.cpan.org #131362]
- update URLs of soundgrab.
0.47 2021-01-29
- Change Specio constraint object's stringification overloading to return the
type name rather than the default Perl object stringification, which gives
you something like
"Specio::Constraint::Parameterized=HASH(0x564d258efb48)". Anonymous are
special cased to return something you can print.
- All types now overload the `eq` comparison operator. Moose expects types to
be comparable in this manner when doing role summation. This fixes a bug
reported by Martin Gruner. GH #18.
0.25026 2020-11-11 03:32:17 -0700
- Handle spaces in Perl and PkgConfig.pm path (gh#47, gh#52)
- Fixed bug where --print-errors and --silence-errors were backwards,
and the default was wrong (gh#53, gh#54, SLAFFAN++)
- Libraries and binaries don't have the .note.GNU-stack section stripped
anymore. Previously, "make strip" would strip that section, which
would sometimes (depending on the toolchain) cause binaries to be
incorrectly tagged as needing an executable stack. This is not a
security issue in itself, but an executable stack makes it easier for
an attacker to turn bugs into exploits, so it should be avoided
whenever possible. Thanks to Xavier Stonestreet for reporting and
finding the cause of the problem.
- Link tests are now performed with a regular file as their
output, instead of /dev/null, which makes them more portable to
old/buggy linkers.
- Bugfixes.
Bump default BUILDLINK_API_DEPENDS to match.
Fix build/test depends.
1.30 2020-10-18
- Support pure Perl builds via the standard incantation of `perl Build.PL
--pureperl_only`. The `--pp` flag is still supported. Implemented by Diab
Jerius. GH @17.
2.4 Sat 26 Sep 2020 18:39:17 BST
- Names of utf8 lexical vars not reported correctly
https://rt.cpan.org/Ticket/Display.html?id=133424
2.5 Sun 27 Sep 2020 13:22:11 BST
- Make 2.4 compatible with older perl.
2.005004 - 2021-03-29
- fix file set in %INC in create_class_with_roles (RT#134845)
2.005003 - 2021-03-21
- silence 'once' warnings from Moo::_Utils
2.005_002 - 2021-01-22
- fix C3 test skipping properly when MRO::Compat not available on perl < 5.10
2.005_001 - 2021-01-20
- fix perl version check in global destruction code
2.005_000 - 2021-01-20
- remove MooseX::Types from developer prereqs
- recommend Sub::Util rather than Sub::Name, since Sub::Util is in core
- fix line numbers when using oo module (perl -Moo)
- adjust some author tests to rely less on external modules
- lower Exporter prereq to any version
- bump Role::Tiny prereq to 2.003004
- refactor and simplify role application code, as allowed by new Role::Tiny
version
- switch to using normal strict+warnings rather than strictures for authors
- remove strictures from recommends, as Moo does not use it anywhere
- remove Task::Weaken prereq, as it served no purpose
- remove MRO::Compat prereq, using it only when user code does
- remove use of Devel::GlobalDestruction in code, only using
Devel::GlobalDestruction::XS when it is available. Devel::GlobalDestruction
is still a dependency on perl < 5.14 to facilitate the installation of the
::XS module.
- Moo now has no mandatory perl version specific dependencies. The version
specific modules are now optional or not used. This means code using Moo can
be fatpacked on a new perl version without special cases, and it will work
on older perl versions.
2.004004 - 2020-11-25
- fix error location test when using core Carp on perl 5.8.9 - 5.10.1
2.004003 - 2020-11-21
- fix extraneous MANIFEST entry
2.004002 - 2020-11-20
- no changes
- releasing as stable
2.004_001 - 2020-07-30
- remove Module::Runtime prerequisite
- internal code cleanups
- added test for conflicts in extensions wrapping 'has' and similar subs
2.002004 - 2021-01-24
- restore requiring base class when using create_class_with_roles, which was
accidentally lost in 2.002_000 (RT#134075)
2.002003 - 2021-01-11
- set dynamic_config to 0 and x_static_install to 1
- fixed filenames in dependents author test to not use colons
- removed MRO::Compat from prereqs as it is no longer used
- marked Class::Method::Modifiers as a hard developer dependency rather than
a recommended developer dependency
- rewrote test for namespace cleaning to not require namespace::clean, and
moved it from an author test to a normal test
- removed namespace::autoclean from developer prereqs
2.002_002 - 2021-01-03
- refactored some internals for subclasses to use
- methods_provided_by will ensure the module is loaded before checking for
methods
2.002_001 - 2020-12-27
- fix tests when Class::Method::Modifiers is not installed
2.002_000 - 2020-12-23
- Refactored create_class_with_roles to not use "composable packages". This
was an optimization for when a role is used in many different uses of
create_class_with_roles, but required an entirely separate implementation,
which did not have fully compatible behavior. It would also result in
methods using modifiers being slower.
- Add documentation for what methods will be composed from a role
(RT#133363)
- Fix backwards compatibility with older versions of Moo when interacting
with Moose.
- Treat modifiers as required methods during create_class_with_roles.
- Fixed methods being no longer composed after they participated in an
allowed conflict.
- Dropped use of MRO::Compat.
- Test cleanups for checking requires during create_class_with_roles.
- Additional tests to confirm internal role application steps can be fully
qualified.
- Fix Pod links to Class::Method::Modifiers
- Tweaks to author tests
- Additional tests for working with older Moo versions
- Additional tests for module loading
- Allow method modifiers to be specified with an array reference of method
names, rather than a list. This now matches how the corresponding
Class::Method::Modifiers subs can be called.
0.430 2020-10-21
- bump List::MoreUtils::XS requirement to 0.430
- release 0.429_001 without any further changes
0.429_001 2020-10-05
- fix failing installation in parallel make (Issue#38)
- infrastructure improvements and tooling updates,
lot's of author tests with according fixes added
- added slide and slideatatime functions wished by SCHWERN
- documentation fixes (PR#21, PR#33, PR#34, RT#126029, RT#132043, RT#132940)
0.430 2020-10-21
- re-release 0.429_002 without further changes
0.429_002 2020-10-05
- fix parts of RT#132043: listcmp misbehave in XS implementation
(thanks to Henrik Pauli for reporting)
0.429_001 2020-10-02
- fix parts of RT#123989: more $a/$b/$_ refcounting bugs
since some bugs are simply reported wrong, dealing with them breaks
more (including running code) than it solves. Way to heavy to feel
better ...
- introduce functions slide and slideatatime (wished by SCHWERN)
- toolchain fixes
- fix RT#133128 - LANG=nb_NO.utf8 related str tests fails (thanks, SJN
for reporting and provide a reasonable fix)
- fix issue#7 (Typo), thanks AMBS for reporting
Changelog:
Simplify types in generics (#663)
Use --profile=check for macro expansion (#671)
Use exported name to prefix enum variants (#672)
Fix path attribute handling in inline submodules (#679)
Fix a stack overflow with some recursive typedefs (#680)
0.78 2021-01-21
[CHANGES]
* Warn on attempts to ->connect to INADDR(6)_LOOPBACK as some OSes
(e.g. OpenBSD) do not allow it
* Various fixes for IO::Async::LoopTests to better support
IO::Async::Loop::UV:
+ Insert some ->loop_once(0) calls between timing tests to allow
libuv to clear its pending queues
+ Allow loops to declare that they cannot perform all-child PID
watch and skip the relevant tests for it if so
* Have $loop->later return a future instance if not passed any code
(RT133240)
[BUGFIXES]
* Accept more error codes for failure to resolve missing hostname
(RT131109)
* Defend against some undef values at global destruction time
(RT132677)
What's New in Pylint 2.8.1?
===========================
Release date: 2021-04-25
..
Put new features and bugfixes here and also in 'doc/whatsnew/2.9.rst'
* Add numversion back (temporarily) in ``__pkginfo__`` because it broke Pylama and revert the unnecessary
``pylint.version`` breaking change.
What's New in Pylint 2.8.0?
===========================
Release date: 2021-04-24
* New refactoring message ``consider-using-with``. This message is emitted if resource-allocating functions or methods of the
standard library (like ``open()`` or ``threading.Lock.acquire()``) that can be used as a context manager are called without
a ``with`` block.
* Resolve false positives on unused variables in decorator functions
* Add new extension ``ConfusingConsecutiveElifChecker``. This optional checker emits a refactoring message (R5601 ``confusing-consecutive-elif``)
if if/elif statements with different indentation levels follow directly one after the other.
* New option ``--output=<file>`` to output result to a file rather than printing to stdout.
* Use a prescriptive message for ``unidiomatic-typecheck``
* Apply ``const-naming-style`` to module constants annotated with
``typing.Final``
* The packaging is now done via setuptools exclusively. ``doc``, ``tests``, ``man``, ``elisp`` and ``Changelog`` are
not packaged anymore - reducing the size of the package by 75%.
* Debian packaging is now (officially) done in https://salsa.debian.org/python-team/packages/pylint.
* The 'doc' extra-require has been removed.
* ``__pkginfo__`` now only contain ``__version__`` (also accessible with ``pylint.__version__``), other meta-information are still
accessible with ``from importlib import metadata;metadata.metadata('pylint')``.
* COPYING has been renamed to LICENSE for standardization.
* Fix false-positive ``used-before-assignment`` in function returns.
* Updated ``astroid`` to 2.5.3
* Add ``consider-using-min-max-builtin`` check for if statement which could be replaced by Python builtin min or max
* Don't auto-enable postponed evaluation of type annotations with Python 3.10
* Update ``astroid`` to 2.5.4
* Add new extension ``TypingChecker``. This optional checker can detect the use of deprecated typing aliases
and can suggest the use of the alternative union syntax where possible.
(For example, 'typing.Dict' can be replaced by 'dict', and 'typing.Unions' by '|', etc.)
Make sure to check the config options if you plan on using it!
* Reactivates old counts in report mode.
* During detection of ``inconsistent-return-statements`` consider that ``assert False`` is a return node.
* Run will not fail if score exactly equals ``config.fail_under``.
* Functions that never returns may declare ``NoReturn`` as type hints, so that
``inconsistent-return-statements`` is not emitted.
* Improved protected access checks to allow access inside class methods
* Fix issue with PEP 585 syntax and the use of ``collections.abc.Set``
* Fix issue that caused class variables annotated with ``typing.ClassVar`` to be
identified as class constants. Now, class variables annotated with
``typing.Final`` are identified as such.
* Continuous integration with read the doc has been added.
* Don't show ``DuplicateBasesError`` for attribute access
* Fix crash when checking ``setup.cfg`` for pylint config when there are non-ascii characters in there
* Allow code flanked in backticks to be skipped by spellchecker
* Allow Python tool directives (for black, flake8, zimports, isort, mypy, bandit, pycharm) at beginning of comments to be skipped by spellchecker
* Fix issue that caused emacs pylint to fail when used with tramp
* Improve check for invalid PEP 585 syntax inside functions
if postponed evaluation of type annotations is enabled
* Improve check for invalid PEP 585 syntax as default function arguments
What's New in astroid 2.5.6?
============================
Release Date: 2021-04-25
* Fix retro-compatibility issues with old version of pylint
What's New in astroid 2.5.5?
============================
Release Date: 2021-04-24
* Fixes the discord link in the project urls of the package.
What's New in astroid 2.5.4?
============================
Release Date: 2021-04-24
* The packaging is now done via setuptools exclusively. ``doc``, ``tests``, and ``Changelog`` are
not packaged anymore - reducing the size of the package greatly.
* Debian packaging is now (officially) done in https://salsa.debian.org/python-team/packages/astroid.
* ``__pkginfo__`` now only contain ``__version__`` (also accessible with ``astroid.__version__``),
other meta-information are still accessible with ``import importlib;metadata.metadata('astroid')``.
* Added inference tip for ``typing.Tuple`` alias
* Fix crash when evaluating ``typing.NamedTuple``
* COPYING was removed in favor of COPYING.LESSER and the latter was renamed to LICENSE to make more apparent
that the code is licensed under LGPLv2 or later.
* Moved from appveyor and travis to Github Actions for continuous integration.
4.75 Wed Dec 30 08:44:55 CET 2020
- removed leftover debug output from treescan.
- added (but not documented) waitid-P_*, FSPICK_*, FSOPEN_*,
FSCONFIG_*, MOUNT_ATTR_* constants.
4.74 Tue Dec 29 16:19:28 CET 2020
- added pidfd_open, pidfd_getfd and pidfd_send_signal functions.
- rework bin/treescan to avoid recursion. This fixes two bugs
where a deep directory traversal or a lot of command line
arguments could cause it to crash.
- added (but not documented) open_tree, AT_* and move_mount flags.
- support defining syscall numbers in gendef0.
4.73 Fri Dec 4 02:08:37 CET 2020
- def0.h was not uptodate, running into musl problems
(reported by B Fraser).
- IO::AIO::splice and IO::AIO::tee didn't properly return
64 bit values.
- added IO::AIO::accept4.
- add various F_SEAL related fcntl constants.
- removed experimental marker for fdlimit functions.
- fiemap now includes the last segment even if it overflows
the end offset, which is arguably the correct behaviour.
0.47 2021-01-01
[BUGFIXES]
* Don't weaken() the waiting future in Future::Mutex as that causes
it to be dropped in some situations
0.46 2020-10-19
[CHANGES]
* Provide AWAIT_CHAIN_CANCEL named method for compatibility with
upcoming Future::AsyncAwait::Awaitable method rename
[BUGFIXES]
* Ensure that Future::Mutex returns proper cloned future instances
when waiting (RT133563)
Add missing build dependency.
1.118 2020-10-21
- Releasing 1.117_001 without further changes
1.117_001 2020-09-25
- fix failing test of dependencies after in 1.116 (from 1.112),
thanks to Dirk Stöcker for reporting via RT#127376 and
Mohammad S Anwar (@manwar) for providing the fix via Github
PR#14
- Fix RT#133368 (PR#15): Fix running tests in parallel submitted
by Kent Fredric (KENTNL) and fix provided by Tom Hukins (@tomhukins)
- Fix RT#125907: spelling error in manpage - thanks to Lucas Kanashiro
for reporting and Graham Knop (@haarg) for kicking me by submitting
PR#17
6.85 2021-03-01
- Test fixes
Missed some tests that were failing.
6.84 2021-03-01
- Time zone fixes
Newest zoneinfo data (tzdata 2021a).
- Language fixes
Corrections and additions to Italian. Patch supplied by Leo
Cacciari (GitHub #33)
2.26 2021-02-01T17:11:00
- Update POD to change RT to github.
- Reformat Makefile.PL.
2.25 2021-01-31T20:29:00
- Merge in clean-up patches from Karen Etheridge, with thanx.
- Add back file t/0. Who deleted it? It is required for t/05.zero.t.
- Update Makefile.PL to switch issue tracker from rt.cpan.org to github.
2.38 2021-01-11 14:35:54 -0700
- Fixed a bug the can cause an array dereference error in pkg-config command line
plugin on MSYS2 / MinGW environment (gh#243)
- Changed the auto-generated module name for Test::Alien, in a way that works
around a probable bug / issue with ExtUtils::CBuilder (gh#243)
Changelog from AN-2021-04-21:
- smake: We now support the idiosyncratic way, GNU make implements ::=
that results in unpredictable behavior with larger makefiles since ::=
creates a different macro type that causes the += operator to behave
different from the usual way introduced in January 1986 by SunPro Make.
+= does not expand the right hand side before appending to an existing
make macro since it has been introduced in January 1986.
+= now expands the right hand side before appending to an existing
::= typed macro. Such a macro is created if the first assignment
to that macro is done via ::=.
Using such a make macro does not expand such macros at use time.
This includes expansions to the left side and the right side of
a colon (:) dependency list and expansions for rule commands.
- smake: The new operators :::= and +:= have been introduced in
order to add an orthogonal (and thus better) counterpart to the
GNU make ::= feature that has become part of the upcoming POSIX
ISSUE 8.
We will propose this as a new POSIX feature soon. In special since
it has been implemented in smake and SunPro Make, see below.
:::= assigns a normal make macro after expanding the right hand side
+:= appends to any maro type after expanding the right hand side
This permits predictable behavior for larger or structured makefile
systems (based on the "include" statement), where it is not possible
to know easily whether a mace macro has been created with a
NAME=value assignment or with a NAME:=value assignment.
Note that the official method in make since 45 years (since make
exists) to prohibit macro expansion is not to use a ::= type macro
but rather to use two '$' signs in the related value of the macro.
- smake: The man page now mentions the new features
- smake: The option -j maxjobs is now supported, but ignored.
This makes smake compatible to the upcoming POSIX ISSUE 8
- smake: added a new function ungetch().
- smake: version bumped to 1.4
Features
t.Fatal snippet
If you trigger completion in a test function, on the line after a
call to a function that returns an error, you will be offered a
completion snippet for:
if err != nil {
t.Fatal(err) }
Experimental
Postfix completion snippets
Postfix completions are a new style of completions that can save
you time by inserting more complex pieces of commonly-written code.
All of the results end with !. Enable this feature by setting
"ui.completion.experimentalPostfixCompletions": true.
The only point of this release is to introduce compatibility with GHCs
back to 7.0 (see https://github.com/feuerbach/tasty/pull/287).
Note, however, that these changes are not merged to the master branch,
and the future releases will only support the GHC/base versions from
the last 5 years, as per our usual policy. To test with even older
GHCs, you'll have to use this particular version of tasty-hunit (or
have the constraint solver pick it for you when testing with older
GHCs).
The source of this release is in the support-old-ghcs branch of the
tasty repository.
Version 1.4.1
* Deduplicate command line options when there is more than one
TestReporter. Expose the function that does that,
uniqueOptionDescriptions.
Version 1.4.0.3
* Fix CPP warning/error macro expansion producing 'defined' has
undefined behavior
Version 1.4.0.2
* Automatically disable the dependency on the clock package when
compiled by ghcjs.
Version 1.4.0.1
* The only point of this release is to introduce compatibility with
GHCs back to 7.0 (see https://github.com/feuerbach/tasty/pull/287).
Note, however, that these changes are not merged to the master
branch, and the future releases will only support the GHC/base
versions from the last 5 years, as per our usual policy. To test
with even older GHCs, you'll have to use this particular version of
tasty (or have the constraint solver pick it for you when testing
with older GHCs).
The source of this release is in the support-old-ghcs branch of the
tasty repository.
Version 1.4
* Change the TreeFold data type to give all functions access to
OptionSet
* Fix a bug where a looping failure message escaped the time out set
for the test
* Fix a bug where pattern changes inside the TestTree weren't
respected
Version 1.3.1
* Add an ability for a test provider to print colorful/formatted
output
Version 1.3
* IsOption has a new method showDefaultValue for customizing how
defaultValues are rendered in the --help output.
* Drop support for GHCs older than 5 years
* Do not install handlers for the signals that dump core
* Export the AnsiTricks type/option
* In addition to a Parser, optionParser and suiteOptionParser now
return a [String] representing warning messages:
- A warning is emitted if an IsOption instance defines multiple
options in the implementation of optionCLParser.
- An warning is emitted if an IsOption instance's optionCLParser
implementation assigns a default value (e.g., with
Options.Applicative.value), as this interferes with tasty's
ability to read environment variable arguments.
1.2.4.1
* Update some test suite dep upper bounds.
1.2.4.0
* Fix a correctness bug with cuckoo hash tables and the new mutate
function introduced in 1.2.3.0.
* Bring test suite into main .cabal file
0.0.13.0
* Ensure that directory listings are returned in sorted order for
reproducibility yesodweb/yesod#1684
0.0.12.0
* Use the Bytes literal on newer GHCs to reduce memory usage during
compilation #36
Version 0.16.1.0 (21 Nov 2020)
* Guard process dependency behind an on by default flag. This allows
one to shrink the dependency tree significantly by turning off the
ability to use bash completion actions.
* Remove bytestring dependency from the test suite.
Version 0.16.0.0 (14 Aug 2020)
* Add Options.Applicative.NonEmpty.some1 function, which parses
options the same as some1 from base, but doesn't cause duplicates in
the usage texts.
* Further improve help text generation in the presence of optional
values when nesting is involved, and many and some when displayed
with a suffix.
* Add "global" options to the usage texts for subcommands. When using
subcommands, a "global options" section can now appear below the
options and commands sections.
Global options are off by default, to enable them, use the
helpShowGlobals modifier.
The noGlobal builder will suppress a single option being displayed
in the global options list.
* Allow the --help option to take the name of a command. Usage without
any arguments is the same, but now, when an argument is given, if it
is the name of a currently reachable command, the help text for that
command will be show.
* Updated dependency bounds.
* Add builder for the all positional parser policy.
* Remove deprecated functions
- nullOption
- execParserMaybe
- customExecParserMaybe
This just does a quick compile/link to actually see if libbacktrace is
present in the current toolchain, not resorting to assumptions based on
Linux distro. In theory, it should work on any system, but I kept the
Linux condition in place.
* Changed the Control.Applicative.Permutations module to only require
Applicative and not Monad. This module is the least restrictive and
works with parsers which are not Monads.
* Added the Control.Monad.Permutations module. This module may be
substantially more efficient for some parsers which are Monads.
* Corrected how permutation parsers intercalate effects and
components; parsing an effect requires that a component immediately
follows or else a parse error will result.
0.3.1.5
* Compatibility with GHC-9.0
- Bump dependencies to allow base >= 4.5 && < 4.16
* Depend on semigroups only with GHC < 8.0
0.3.1.4
* Compatibility with GHC-8.10
- Bump dependencies to allow base >= 4.5 && < 4.15
1.2.4.2
* Mask exceptions in Acquire allocation action
1.2.4.1
* Document risk of using forkIO within a ResourceT #441
1.2.4
* Add allocate_ #437
1.2.3
* Support unliftio-core 0.2.0.0
Fix a bug in which give (and possibly reify, reifyNat, and
reifySymbol) could be unsoundly inlined by GHC 8.10 or older to
produce incorrect runtime results.
1.1.5
* Fix build failures on GHC 8.4 and 8.6
* Add support for safe Haskell
- Specifically, this marks the Control.Monad.Trans.Compose module as
Trustworthy
- The change in 1.1.4 to use GeneralizedNewtypeDeriving meant that
the Control.Monad.Trans.Compose module was no longer inferred as
safe
* Restore Traversable instance removed by mistake in 1.1.4
1.1.4 (Blacklisted)
* Unintentional removal of Traversable instance for ComposeT
- This missing instance is restored in 1.1.5
- This is the reason why the 1.1.4 release is blacklisted
* Fix MonadFail-related code to work for GHCJS
* The MonadRWS instance for ComposeT has a more flexible constraint
- The constraint is now MonadReader r (f (g m)), MonadWriter w (f (g
m)), MonadState s (f (g m)) instead of MonadRWS r w s (f g m)
- This loosening of the constraint is backwards-compatible
0.9.0.4
* Support GHC 9.0.1
0.9.0.3
* Support GHC 8.10
* Drop support for GHC 8.4
* Hint can now be used concurrently from multiple threads on GHC 8.10+
Version 0.11
* Remove deprecated getCursorPosition0. (Use getCursorPosition
instead.)
* On Unix-like operating systems, the temporary turning off of echoing
is moved from getReportedCursorPosition to hGetCursorPositon.
* On Unix-like operating systems, fix a bug in getCursorPosition and
hGetCursorPosition, where the console input stream was was not
always clear before the cursor position was emitted into it.
Version 0.10.3
* Add getCursorPosition as a synonym of getCursorPosition0 and
deprecate the latter.
QuickCheck 2.14.2 (released 2020-11-14)
* Add Arbitrary instances for Tree (thanks to Oleg Grenrus)
* GHC 9.0 compatibility (thanks to Vilem-Benjamin Liepelt)
QuickCheck 2.14.1 (released 2020-07-05)
* Compatibility with random >= 1.2.
QuickCheck 2.14 (released 2020-03-28)
* QuickCheck is now much faster at generating test data!
As a result, many properties can now be tested a lot faster; the
examples distributed with QuickCheck run about twice as fast as
before, for example. Of course, your mileage may vary.
The reason for this is that there are now specialised versions of
the 'choose' combinator for various types. These are: chooseInt,
chooseInteger, chooseBoundedIntegral, and chooseEnum. These
combinators are identical to 'choose' but much faster. All
QuickCheck combinators, with the exception of 'choose' itself, use
the new combinators behind the scenes.
You should see a speedup without doing anything, but to get a
further speedup, consider replacing any uses of 'choose' in your own
generators with the new combinators.
We are hoping that future releases of the 'random' library may speed
up 'choose', in which case these combinators may no longer be
needed.
Thanks to Oleg Grenrus for suggesting to bypass 'choose' for random
number generation, and providing the appropriate primitive in his
'splitmix' random number library.
* Smaller changes and bugfixes:
- RecursivelyShrink and GSubterms are exported from
Test.QuickCheck.Test (thanks to Tom Mortiboy).
- Don't generate invalid unicode characters (thanks to Boris
Stepanov).
- When a call to 'within' fails, include the duration of the timeout
in the failure report (thanks to William Rusnack).
- In Gen, avoid splitting the seed in the implementation of >>, *>
and <* (thanks to David Feuer).
- Fix a couple of bugs with shrinking of floating-point numbers.
- Export functionMapWith, functionEitherWith and functionPairWith
from Test.QuickCheck.Function (thanks to Oleg Grenrus).
- Remove redundant RealFloat constraint from Arbitrary/CoArbitrary
instances for Complex (thanks to Bodigrim).
0.5.3 (8 April 2021)
* StatefulGen instances for RandT
* Addition of RandGen
* Additioon of withRandGen and withRandGen_
0.5.2 (24 June 2020)
* Support for random-1.2.
* Drop support for GHC 7.4, test with GHC 8.10.
0.5.1.2 (18 Jan 2020)
* Better error message when total probability sum is negative in
fromListMay, weightedMay, weighted
0.2.2.1 -- 2020-10-16
* Relaxed upper bound for hslua, allow hslua-1.3.*.
0.2.2 -- 2020-08-15
* Relaxed upper bound for hslua, allow hslua-1.2.*.
* Improved documentation of internal types.
* Use tasty-lua for unit tests.
* Update CI to test with all GHC versions.
Changes in 0.11.2 [2020.09.30]
* This coincides with the base-compat-0.11.2 release. Refer to the
base-compat changelog for more details.
Changes in 0.11.1 [2020.01.27]
* This coincides with the base-compat-0.11.1 release. Refer to the
base-compat changelog for more details.
Changes in 0.8.4 [2020.12.09]
* Backport the Ord instances for TyCon and TypeRep to be available on
GHC 7.0.
Changes in 0.8.3 [2020.09.30]
* Backport new instances from GHC 9.0/base-4.15
- MonadFix and MonadZip instances for Complex
- Ix instances for tuples of size 6 through 15
Changes in 0.8.2 [2020.01.27]
* Backport new instances from GHC 8.10/base-4.14
- Bits, Bounded, Enum, FiniteBits, Floating, Fractional, Integral,
Ix, Real, RealFrac, RealFloat and Storable instances for
Data.Ord.Down
- Functor, Applicative, Monad, Alternative, and MonadPlus instances
for Kleisli
- Functor, Applicative, and Monad instances for (,,) a b and (,,,) a
b c
- Data instances for WrappedArrow and WrappedMonad
- Data and IsList instances for ZipList
- TestEquality instance for Compose
* Backport the Typeable (() :: Constraint) instance to be available on
GHC 7.8.
- Fixing comments on batch report, now compatible with PHP 7
- Try to fix recurring problems with test runs hanging when the last
test requires a rerun due to a knownbug
- #82 trying to support Python 3.9
* Add FunctorClassesDefault, an adapter newtype suitable for
DerivingVia, to Data.Functor.Classes.Generic.
* Fix a bug in which readsPrec1Default/liftReadsPrecDefault would
parse empty data types too strictly.
* Remove instances for Data.Semigroup.Option, which is deprecated as
of base-4.15.0.0.
* Allow building with template-haskell-2.17.0.0 (GHC 9.0).
* Fix a bug in which deriveAll1 would needlessly reject data types
whose last type parameter appears as an oversaturated argument to a
type family.
* Declare Data.UUID.Types module "Trustworthy" with respect to
SafeHaskell.
* Use more compact heap object representation which saves 16 bytes on
64bit platforms.
* Add toWords64/fromWords64 functions
* Drop support for GHC < 7.
* Add support for random-1.2, i.e. idiomatic Random and added Uniform
instances.
* Breaking change which mostly maintains backwards compatibility, see
"Breaking Changes" below.
* Support for monadic generators e.g. mwc-random.
* Monadic adapters for pure generators (providing a uniform monadic
interface to pure and monadic generators).
* Faster in all cases except one by more than x18 (N.B. x18 not 18%)
and some cases (depending on the type) faster by more than x1000
* Passes a large number of random number test suites:
- dieharder
- TestU01 (SmallCrush, Crush, BigCrush)
- PractRand
- gjrand
- See https://github.com/tweag/random-quality for details on how to
do this yourself.
* Better quality split as judged by these tests:
https://www.cambridge.org/core/journals/journal-of-functional-programming/article/evaluation-of-splittable-pseudorandom-generators/3EBAA9F14939C5BB5560E32D1A132637
Again see https://github.com/tweag/random-quality for details on how
to do this yourself.
* Unbiased generation of ranges.
* Updated tests and benchmarks.
* Continuous integration.
Breaking Changes
Version 1.2.0 introduces these breaking changes:
* requires base >= 4.8 (GHC-7.10)
* StdGen is no longer an instance of Read
* randomIO and randomRIO were extracted from the Random class into
separate functions
In addition, there may be import clashes with new functions,
e.g. uniform and uniformR.
Deprecations
Version 1.2.0 introduces genWord64, genWord32 and similar methods to
the RandomGen class. The significantly slower method next and its
companion genRange are now deprecated.
0.1.0.3
* Fix oops bugs in 0.1.0.2
- It's lowercase windows.h. I blame Microsoft docs for using capital
case Windows.h in the
docs. https://docs.microsoft.com/en-us/windows/win32/api/processthreadsapi/nf-processthreadsapi-getprocessid
- accidental shiftL vs shiftR mixup for 32-bit generator
initialization. Doesn't affect Linux.
0.1.0.2
* Drop time dependency in favour of handcoded initialization
- On Unix platforms we use /dev/urandom if it exists, otherwise use
gettimeofday, clock and getpid.
- On Windows we use GetCurrentProcessID, GetCurrentThreadId(),
GetTickCount, GetSystemTime and QueryPerformanceCounter.
- On GHCJS use Math.random()
- Using time is a fallback option (e.g. for Hugs).
0.1.0.1
* Add INLINEABLE pragmas to bitmaskWithRejection* functions
* Support GHC-9.0
0.1
* Drop random dependency
unconditionally. https://github.com/phadej/splitmix/issues/34
0.0.5
* Add nextInteger
* Use smaller range in bitmaskWithRejection32 and 64, when upper bound
is 2^n - 1. This changes generated values when they were on the
boundary.
0.0.4
* Add bitmaskWithRejection32' and bitmaskWithRejection64' which
generate numbers in closed range [0, n]. Unticked variants generate
in closed-open range [0, n).
0.2.13.0
* Add HashMap.compose. Thanks Alexandre Esteves.
0.2.12.0
* Add HashMap.isSubmapOf[By] and HashSet.isSubsetOf. Thanks Sven
Keidel. (#282)
* Expose internal modules. (#283)
* Documentation improvements in Data.HashSet, including a
beginner-friendly introduction. Thanks Matt Renaud. (#267)
* HashMap.alterF: Skip key deletion for absent keys. (#288)
* Remove custom unsafeShift{L,R} definitions. (#281)
* Various other documentation improvements.
0.2.11.0
* Add HashMap.findWithDefault (soft-deprecates
HashMap.lookupDefault). Thanks, Matt Renaud.
* Add HashMap.fromListWithKey. Thanks, Josef Svenningsson.
* Add more folding functions and use them in Foldable
instances. Thanks, David Feuer.
* Add HashMap.!?, a flipped version of lookup. Thanks, Matt Renaud.
* Add a Bifoldable instance for HashMap. Thanks, Joseph Sible.
* Add a HasCallStack constraint to (!). Thanks, Roman Cheplyaka.
Bug fixes
* Fix a space leak affecting updates on keys with hash
collisions. Thanks, Neil Mitchell. (#254)
* Get rid of some silly thunks that could be left lying
around. (#232). Thanks, David Feuer.
Other changes
* Speed up the Hashable instances for HashMap and HashSet. Thanks,
Edward Amsden.
* Remove a dependency cycle hack from the benchmark suite. Thanks,
Andrew Martin.
* Improve documentation. Thanks, Tristan McLeay, Li-yao Xia, Gareth
Smith, Simon Jakobi, Sergey Vinokurov, and likely others.
0.4.2.0 -- 2020-12-30
* Explicitly mark modules as Safe (or Trustworthy for GHC versions
prior to 8.4).
0.4.1.0 -- 2020-12-09
* Fix a bug in which normalizeDec would give incorrect kind
annotations to type variables in quoted Decs. normalizeDec now
leaves the kinds of type variable binders alone.
0.4.0.0 -- 2020-09-29
* Adapt to the TyVarBndr data type gaining a new flag type parameter
(in template-haskell-2.17.0.0) to represent its specificity:
* Introduce a new Language.Haskell.TH.Datatype.TyVarBndr module that
defines TyVarBndr_, a backwards-compatible type synonym for
TyVarBndr, as well as backporting TyVarBndrSpec, TyVarBndrUnit, and
Specificity. This module also defines other useful functions for
constructing and manipulating TyVarBndrs.
* The types in Language.Haskell.TH.Datatype now use TyVarBndr_,
TyVarBndrUnit, and TyVarBndrSpec where appropriate. Technically,
this is not a breaking change, since all three are simple type
synonyms around TyVarBndr, but it is likely that you will need to
update your th-abstraction-using code anyway if it involves a
TyVarBndr-consuming function.
0.3.2.0 -- 2020-02-06
* Support substituting into and extracting free variables from
ForallVisTs on template-haskell-2.16.0.0 (GHC 8.10) or later.
* Fix a bug in which freeVariables could report duplicate kind
variables when they occur in the kinds of the type variable binders
in a ForallT.
* Fix a bug in which resolveInfixT would not resolve UInfixTs
occurring in the kinds of type variable binders in a ForallT.
* Fix a bug in which the TypeSubstitution ConstructorInfo instance
would not detect free kind variables in the constructorVars.
v1.0
Released on 2020-07-18, Nelson Mandela International Day.
Added
* intercalate for DList (#43, Jacob Leach)
* Traversable instance for DList (#45, Veronika Romashkina)
* Data.DList.Internal for the DList implementation, Data.DList.Unsafe
for exporting the DList constructor UnsafeDList and record label
unsafeApplyDList (#55, #59)
* Data.DList.DNonEmpty (#60)
* GitHub Action for uploading a release (#74)
* dlist-bench, a benchmark package (#71)
Changed
* stimes for DList defined with stimesMonoid (#46, Janek Spaderna)
* Type of tail: DList a -> DList a to DList a -> [a] (#69)
* GitHub Action for continuous integration testing to replace Travis-CI (#47, #50)
* GHC warning and error improvements (#72, #73)
* Improved documentation (#55, #70, #76, #77)
Removed
* list :: b -> (a -> DList a -> b) -> DList a -> b (#69)
v0.8.0.8
Released on 2020-04-02, World Autism Awareness Day.
Added
* toList in the Foldable instance for DList (#36, Ryan Scott)
Changed
* QuickCheck upper bound: 2.14 to 2.15 (a7ea60d)
Fixed
* Documented time complexity of head for DList (#35, Simon Jakobi)
Changes in 0.11.2 [2020.09.30]
* Sync with base-4.15/GHC 9.0
* Backport singleton to Data.List and Data.List.NonEmpty
* Backport hGetContents', getContents', and readFile' added to System.IO
Changes in 0.11.1 [2020.01.27]
* Sync with base-4.14/GHC 8.10
* Backport isResourceVanishedError, resourceVanishedErrorType, and
isResourceVanishedErrorType to System.IO.Error.Compat.
Changes in version 0.7.1.0
* Introduce convenience class MonadPrim and MonadPrimBase.
* Add PrimMonad and PrimBase instances for Lazy.ST (GHC >=
8.2). thanks to Avi Dessauer (@Avi-D-coder) for this first
contribution
* Add freezeByteArray and freezePrimArray.
* Add compareByteArrays.
* Add shrinkMutableByteArray.
* Add Eq instances for MutableByteArray and MutablePrimArray. by
Andrew Martin
* Add functions for manipulating pinned Prim Arrays by Andrew Martin
* Add copyPtrToMutableByteArray.
* Add NFData instances for ByteArray, MutableByteArray, PrimArray and
MutablePrimArray. by Callan McGill
* Add shrinkSmallMutableArray.
* Add clonePrimArray and cloneMutablePrimArray.
* Add cloneMutableByteArray and cloneByteArray.
* Add Prim instances for WordPtr and IntPtr.
* Add NFData instances for Array and SmallArray. by Callan McGill
* Add copyByteArrayToPtr and copyMutableByteArrayToPtr.
* Export arrayFromList and arrayFromListN.
Changes in version 0.7.0.1
* Allow building with GHC 8.12. Thanks Ryan GL Scott for this and
every compat patch over time.
* Fix#121: the -i flag produces an .info file even if
the %expect pragma is violated
* Fix#131: qualify uses of Prelude functions in generated code
* Fix#161: drop fewer parse items when generating an .info file
* Introduce the %shift directive to resolve shift/reduce
conflicts explicitly, useful in conjunction with %expect 0
* Remove the deprecated build configuration flag 'small_base'
Changes in 3.2.6:
Support for the GHC 9.2.
The array access primops now use the fixed-sized numeric types
corresponding to the width of the data accessed. Additionally, the
primops to convert to and from fixed-sized numeric types have been
given new names.
9.2 isn't cut yet, so these changes are somewhat
speculative. Unfortunately, GHC must use a released version of
Alex (and Happy) at all times until further changes have been
made, so we must make the release to actually implement these
changes in GHC.
If the final GHC 9.2 ends up being different, this release will be
marked broken to make it less likely people use it by accident.
5.0.9:
Improve 3.9 IS_OP and CONTAINS_OP operand formatting
Correct disassembly of 3.9 bytecode from other bytecode
Accept 3.9.3 and 3.9.4 as a valid 3.9 version
Accept 3.8.9 as a valid 3.8 version
0.15.1 (2021-04-22)
Hotfix for errors while closing event loops while replacing them.
0.15.0 (2021-04-19)
Add support for Python 3.9
Abandon support for Python 3.5. If you still require support for Python 3.5, please use pytest-asyncio v0.14 or earlier.
Set unused_tcp_port_factory fixture scope to 'session'.
Properly close event loops when replacing them.
Updated Libraries
Asio:
Added ip::scope_id_type type alias.
Added ip::port_type type alias.
Added std::hash specialisations for IP addresses.
Added std::hash specialisations for ip::basic_endpoint<>.
Refactored SFINAE usage to improve compile times.
Added friendship support to customisation points, and made most customisations private.
Changed any_io_executor to a "strong typedef"-style class.
Fixed experimental::as_single to work with handler hook deprecation.
Ensured pthread condition variable attributes are cleaned up on all platforms.
Clarified thread safety notes on sockets and descriptors.
Ensured errno is not overwritten if socket() fails on macOS/FreeBSD.
Fixed work tracking for io_context and thread_pool executors when move-assigned.
Ensured internal call_stack objects are accessed only from implementation files.
Fixed I/O object move-assignment to ensure the executor is left in a valid state.
Fixed detection of compiler support for defaulted template argument on functions with MSVC.
Prevented the blocking.always property from being used with strand<>, as it did not produce the correct semantics.
Removed deprecated file asio/impl/src.cpp.
Atomic:
Fixed compilation with MSVC for ARM.
Beast:
This update brings minor bug fixes and revamped CI reporting.
We'd love to know how you or your company use Beast, consider adding an entry to the Companies and Individuals Using Beast list.
See the full Release Notes for a complete list of changes.
Bind:
Add support for using the standard C++11 placeholders with boost::bind.
Update boost::apply to use variadic templates and perfect forwarding.
Container:
Added [[no-discard]] attribute in all containers to catch bugs related to unused return values.
Replaced default standard exception classes with Boost.Container own classes, reducing considerably the included files overhead. Example: in MSVC 19 boost/container/vector.hpp preprocessed file size reduces from 1,5MB to 930KB. If you still want to use standard exception classes, you can define BOOST_CONTAINER_USE_STD_EXCEPTIONS before using any Boost.Container class.
Fixed bugs/issues:
GitHub 102: "flat_map::insert ambiguous with initializer list & pairs that need to convert".
GitHub 139: "flat_map merge and iterators".
GitHub 141: "small_vector does not propagate no throw properties of move operation of contained type".
GitHub 164: "Compile error when using pmr::map with a std::pair; works when using a std::tuple".
GitHub 171: "deque::clear() uses undefined behaviour".
Core:
Add implicit conversion between compatible reference wrappers.
Add boost/core/cmath.hpp, a portable implementation of the floating point classification functions from <cmath>.
Add boost/core/bit.hpp, a portable implementation of the C++20 standard header <bit>.
Fix BOOST_TEST_EQ, BOOST_TEST_NE for character types under C++20.
Revise allocator access utilities (now support VS2013, and no workarounds use allocator_traits.)
DLL:
BREAKING CHANGE: boost::dll::import was renamed to boost::dll::import_symbol to avoid collision with C++20 import keyword.
Updated docs, including FAQ section.
Filesystem:
Updated compatibility with WASI platform.
Fixed an exception being thrown by path::remove_filename if the path is "////".
Fixed create_directories disregarding errors from file status query operations issued internally. This could result in incorrect error codes returned by create_directories.
GIL:
BREAKING: In next release, we are going to drop support for GCC 5. We will also change the required minimum C++ version from C++11 to C++14.
Intrusive:
Reduced compile-time dependencies:
linear_slist_algorithms use a simple node_ptr instead of std::pair on return.
list/slist use operator </operator == instead of std::equal_to/std::less.
Fixed GitHub 54: set.rbegin() looks like O(log(N))
Interprocess:
Added wchar_t API support for named resources in operating systems that offer native wide character API (e.g. Windows). The following classes were updated with wchar_t name support:
file_mapping
managed_mapped_file
managed_shared_memory
managed_windows_shared_memory
shared_memory_object
windows_shared_memory_object
file_lock
named_condition
named_condition_any
named_mutex
named_recursive_mutex
named_semaphore
named_sharable_mutex
named_upgradable_mutex
message_queue
Added BOOST_INTERPROCESS_FORCE_NATIVE_EMULATION macro option to disable the use of generic emulation code for process shared synchronization primitives instead of native POSIX or Winapi functions.
Fixed bugs:
GitHub 76 ("Cygwin compilation errors").
GitHub 83 ("Add BOOST_INTERPROCESS_FORCE_NATIVE_EMULATION option").
GitHub 92 ("bufferstream: Correct MSVC compilation warning").
GitHub 106 ("Use fallocate on truncate_file").
GitHub 120 ("segment_manager customization").
GitHub 122 ("Mark constructors/assignment/swap noexcept where possible").
GitHub 126 ("_ReadWriteBarrier is deprecated warning when compiling with clang-cl.exe").
JSON:
Security Report from Bishop Fox.
Refactored value_from implementation; user customizations are now always preferred over library-provided overloads.
Fix imprecise parsing for some floating point numbers.
Fix link errors in standalone mode, when used alongside Boost.
Fix Boost.Build builds on GCC 4.8.
LexicalCast:
Dropped dependency on Boost.Math
Fixes for the CI and coverage
Log:
Bug fixes:
Corrected a formal race condition in the thread-safe log record queue implementation used in the unbounded_fifo_queue policy of the asynchronous_sink frontend. The race could be detected by TSAN, but it was not a real problem on most current CPU architectures and compilers.
When copying a logger with a channel attribute (e.g. channel_logger), the channel attribute is now deep-copied from the original logger. This means that modifying the channel in the new logger will no longer affect the original logger.
Replaced the use of std::allocator<void> in template parameters with a new tag type boost::log::use_std_allocator to silence libc++ warnings about the former being deprecated in C++17 and later. The tag indicates that the instantiated template should be using a specialization of std::allocator internally to allocate dynamic memory, so the change has no functional effect.
Boost.Log no longer defines __MSVCRT_VERSION__ macro on MinGW and MinGW-w64. Defining this macro caused incompatibility with UCRT, which is available as an option in recent MinGW-w64.
See changelog for more details.
Math:
Breaking Change: C++03 support is now removed, a C++11 or later conformant compiler is now required to use this library.
Added Z-test.
Added execution policy support to univariate and bivariate statistics: enables parallel execution (requires C++17 and <execution>).
Big update/improvement on CI testing.
Bivariate statistics now have integer support.
T-Test now has integer support.
Linear regression now has integer support.
Correct PDF of the beta distribution at the endpoints.
Correct use of Stirling's approximation in certain multiprecision cases.
Eliminate -Wimplicit-int-float-conversion on clang.
Fix some constexpr issues in quaternion/octonion.
Minor performance fix to tanh_sinh integration.
Update hypergeometric functions internal scaling to allow for 64-bit (long long) exponents with multiprecision types.
Move:
Git Issue 35: "New nothrow move traits are incomplete".
Multiprecision:
BREAKING CHANGE: Massive refactoring and code simplification makes C++11 an absolute requirement.
Use BOOST_TRY/CATCH in headers so code can be used in exception-free environments.
Correct corner case in pow.
Correct exception type thrown to match docs in lsb/msb.
Allow moves and operators between related but different types (ie types with the same allocator).
Nowide:
Fix discarding of characters for text file streams when seek or sync functions are used and newlines were converted (e.g. on Windows)
Fix potential use-after-free bug when reusing (via open) a file stream that has been closed
Fix potential invalid memory access when using a file stream that has been moved or swapped to where the source had a put-back character or was set to unbuffered mode
Fix compatibility with ucrt runtime by not defining __MSVCRT_VERSION__
Known issues: Read performance for text files is degraded. Binary files and writing is unaffected.
Optional:
Fixed MSVC warning C4702.
Outcome:
Announcements:
BREAKING CHANGE: After a year and three major Boost releases announcing this upcoming change, this is the FIRST RELEASE of the v2.2 branch. This branch has a number of major breaking changes to the Outcome v2.1 which shipped in Boost 1.75 and earlier, see the documentation for details.
Enhancements:
VS2019.8 changed how to enable Coroutines, which caused Outcome to not compile on that compiler.
If on C++ 20, we now use C++ 20 [[likely]] instead of compiler-specific markup to indicate when TRY has likely success or failure.
BREAKING CHANGE: Previously the value of spare_storage(const basic_result|basic_outcome *) noexcept was not propagated over BOOST_OUTCOME_TRY, which causes things like stack backtraces captured at the point of construction of an errored result to get dropped at every TRY point. This has been fixed by adding an optional spare_storage to success_type<T> and failure_type<T>, as well as to auto success(T &&, ...) and auto failure(T &&, ...).
Bug fixes:
Boost.Outcome should now compile with BOOST_NO_EXCEPTIONS defined. Thanks to Emil, maintainer of Boost.Exception, making a change for me, Boost.Outcome should now compile with C++ exceptions globally disabled. You won't be able to use boost::exception_ptr as it can't be included if C++ exceptions are globally disabled.
BREAKING CHANGE 244 It came as a shock to learn that BOOST_OUTCOME_TRY had been broken since the inception of this library for certain corner case code, which required a breaking change in how TRY works. See the changelog in the documentation for more detail.
Parameter:
Added a workaround for MSVC 2015 code miscompilation when an rvalue was used as a default value of a named parameter of a function.
PFR:
Fixed reflection of types with missing const in SFINAE expressions in template constructor. Now reflection of aggregates with std::optional<std::chrono::duration<???>> fields works on any Standard Library implementation, even if LWG3050 is not fixed.
Comparison functions are now constexpr
Fixed numerous typos in docs
PolyCollection:
Worked around GCC bug affecting GCC versions 9.3-10.2.
Predef:
Version 1.13
Add ARCH_PPC_64 predef.
Fix ARCH_WORD_BITS* redefinition warnings/errors.
Add ARCH_E2K, Elbrus 2000, architecture from Konstantin Ivlev.
Fix not handling recent C++ version that go above 10.x version.
Version 1.12
Switch to using the endian.h header on OpenBSD. (Brad Smith)
Fix not handling recent versions of stdcxx that go above version 9.
Fix including sub-BSD OS headers directly causing redef warnings.
Add CI testing of direct inclusion of all headers.
Add CI testing on FreeBSD for clang and gcc.
Add WORD_BITS set of predefs to detect the architecture word size. Initial implementation inspired by submission from Mikhail Komarov.
Add CI testing for Cygwin 32 and 64.
PropertyTree:
Property Tree has a new maintainer.
Fix usage of deprecated Boost.Bind features.
Link to Wikipedia fixed.
Use BOOST_OVERRIDE to fix GCC warnings.
Add extensive CI testing.
Regex:
Regex is now header only except in C++03 mode.
Support for C++03 is now deprecated.
The library can now be used "standalone" without the rest of Boost being present.
Stacktrace:
Better support for pre-C++11 builds on clang.
Bigger warning for do not use boost::stacktrace::stacktrace in signal handlers.
TypeTraits:
Fix for Visual Studio managed code projects.
Variant2:
Improved generated code for the double buffered case.
Updated Tools
Build:
Includes release of B2 version 4.4.1.
6.10.0 - 2021-04-17
This release teaches the Ghostwriter to read parameter types from
Sphinx, Google, or Numpy-style structured docstrings, and improves
some related heuristics about how to test scientific and numerical
programs.
6.9.2 - 2021-04-15
This release improves the Ghostwriter’s handling of exceptions, by
reading :raises ...: entries in function docstrings and ensuring
that we don’t suppresss the error raised by test assertions.
6.9.1 - 2021-04-12
This patch updates our autoformatting tools, improving our code
style without any API changes.
6.9.0 - 2021-04-11
This release teaches from_type() how to see through typing.Annotated.
Thanks to Vytautas Strimaitis for reporting and fixing issue #2919!
6.8.12 - 2021-04-11
If rich is installed, the hypothesis write command will use it to
syntax-highlight the Ghostwritten code.
6.8.11 - 2021-04-11
This patch improves an error message from from_type() when builds()
would be more suitable (issue #2930).
6.8.10 - 2021-04-11
This patch updates the type annotations for arrays() to reflect
that shape: SearchStrategy[int] is supported.
6.8.9 - 2021-04-07
This patch fixes from_type() with abstract types which have either
required but non-type-annotated arguments to __init__, or where
from_type() can handle some concrete subclasses but not others.
6.8.8 - 2021-04-07
This patch teaches hypothesis write to check for possible roundtrips
in several more cases, such as by looking for an inverse in the
module which defines the function to test.
6.8.7 - 2021-04-07
This patch adds a more helpful error message if you try to call
sampled_from() on an Enum which has no members, but does have
dataclass()-style annotations (issue #2923).
Changes from Ant 1.10.9 TO Ant 1.10.10
======================================
Fixed bugs:
-----------
* SCP (with sftp=true) task would fail if fetching file located in root directory
Bugzilla Report 64742
* javac task would fail if the arguments file it (internally) created didn't quote
the # character. This has now been fixed.
Bugzilla Reports 64912, 64790
* made sure LegacyXmlResultFormatter encodes characters illegal in
XML the same way JUnit5's built-in formatter would.
Bugzilla Report 65030
* LegacyXmlResultFormatter no longer double-encodes <>& in system-err
and system-out
Bugzilla Report 63436
* Fixes a bug in junitlauncher task's legacy-xml formatter, where the testcase
representing a @Parameterized JUnit4 test wasn't being reported in the XML.
Bugzilla Report 64952
* Fixes a bug where the ant-testutil-sources.jar that gets published to Maven
central repository didn't contain any source files.
Bugzilla Report 65110
* The <http> condition didn't follow redirects from http to https.
Bugzilla Report 65105
* ZipOutputStream now overrides write(int) in order to make sure
single byte writes get the same treatment as array writes.
Github Pull Request #145
* Fixes a potential deadlock in junitlauncher task when using legacy-xml
reporter.
Bugzilla Report 64733
Other changes:
--------------
* javaversion condition now has a new "atmost" attribute. See the javaversion
manual for more details
* The "listener" nested element of the "junitlauncher" task now has a new
"useLegacyReportingName" attribute which can be used to control the test
identifiers names that get reported by the listener. See the junitlauncher
manual for more details.
Note that this change also introduces a new "setUseLegacyReportingName" method
on the org.apache.tools.ant.taskdefs.optional.junitlauncher.TestResultFormatter
interface. This will break backward compatibility with any of your custom
result formatters which implemented this interface and such implementations
are now expected to implement this new method.
* a new attribute preserveduplicates allows <resourcelist> to return
the same resource multiple times when set to true.
Bugzilla Report 64854
* a new attribute filterbeforeconcat in <concat> can be used to
decide whether the filterchain should be applied to the
concatenated content (the default) or each nested resource
individually before concatenating them.
Bugzilla Report 64855
* the ssh tasks now share a new nested element additionalConfig that
can be used to set config values for the jsch Session used by the
task.
Bugzilla Report 65089
* added new discardOutput and discardError properties to redirector
and the exec, apply and java tasks which can be used to completely
discard any (error) output. This is a platform independent
alternative to directiong output to any kind of null device.
- Less noisy output: Only show number of unpushed commits in output if
there are more than 0.
- Add experimental command `mob squash-wip` to squash any WIP commits in
the wip branch into a following manual commit using `git rebase
--interactive` with `mob` as the temporary `GIT_EDITOR`.
- The order of the latest commit is now reversed, the latest one is
shown last.
- Add experimental configuration option `MOB_WIP_BRANCH_PREFIX` to
configure the `mob/` prefix to some other value.
version 3.35.5:
Fix defects in the new ALTER TABLE DROP COLUMN feature that could corrupt the database file.
Fix an obscure query optimizer problem that might cause an incorrect query result.
Changes from 5.2.6 to 5.2.7 (March 2021)
****************************************
The main developers this time were: SiegeLord, Peter Hull, Elias
Pschernig, Aldrik Ramaekers, Andreas Rönnquist.
Build system:
• Allow generating projects with a suffix (lorry-lee).
• Fix build under Clang-CL in Visual Studio.
Core:
• Avoid some undefined behavior errors.
• Return key modifiers in ‘ALLEGRO_EVENT_KEY_UP’ and
‘ALLEGRO_EVENT_KEY_DOWN’.
• Allow calling ‘al_map_*’ color functions before Allegro is
initialized.
• Allow minimum bitmap size to be something other than 16 on
non-Android platforms (controlled via ‘allegro5.cfg’).
• Add ‘al_get_monitor_refresh_rate’ (only implemented on Windows for
now).
Graphics:
• Fix ‘ALLEGRO_KEEP_INDEX’ flag for bitmaps.
• Add ‘ALLEGRO_OPENGL_CORE_PROFILE’ display flag.
Emscripten:
• The experimental Emscripten support (via the SDL backend) is now
documented in ‘README_sdl.txt’.
OSX:
• Move more Cocoa operations to the main thread.
• Explicitly link CoreVideo to fix the static build.
Windows:
• Issue #1125: Speed up OpenGL extension detection (Tobias Scheuer).
• Use Unicode APIs when enumerating joysticks.
• Use ‘WM_DEVICECHANGE’ rather than polling to detect joystick
hotlugging, reducing input drops and lags (Todd Cope).
• Fix joystick polling period.
• Restore WinXP compatibility by using slightly older API when
loading shared libraries (Julian Smythe).
• Fix build with HLSL disabled (Julian Smythe).
• Raise DirectInput ‘MAX_JOYSTICKS’ to 32 and ‘DEVICE_BUFFER_SIZE’ to
128.
SDL:
• Issue #1224: Fix bug in SDL voice driver.
Audio addon:
• Allows playing sounds in reverse by specifying a negative speed.
Acodec addon:
• Fix edge-case looping in Ogg Vorbis stream (Cody Licorish)
Audio addon:
• Use more sensible values for PulseAudio’s playback buffer,
potentially resolving some crashes and high CPU usage.
Native Dialog Addon:
• Migrate from GTK2 to GTK3. Sadly, we lose menu icons as GTK3
dropped support for them.
TTF addon:
• Allow initializing TTF addon before the Font addon.
• Shut-down the TTF addon automatically in ‘al_uninstall_system’.
PhysFS addon:
• Fix handling of native path separators.
• Stop using deprecated PhysFS API.
Primitives addon:
• Fix segfault in ‘al_draw_ribbon’ when ‘num_segments > 128’ (Rodolfo
Borges).
• Issue 1215: Correctly handle small scales when determining
subdivision level for high level primitives (Robin Heydon).
Documentation:
• Fix LaTeX errors in the generation of the reference manual PDF.
• Add links to examples into the reference manual.
• Allow pressing ‘S’ to focus the search bar in the docs.
• Assorted documentation improvements.
Misc:
• Add a security policy and an associated private security mailing
list - allegro-security@lists.liballeg.org.
• Add Emscripten-powered examples to
https://liballeg.org/examples_demos.html.
Examples:
• ‘ex_audio_simple’ now displays instructions and supports
bidirectional looping.
• Add default files to some audio examples.
CLI::Osprey is a module to assist in writing commandline applications
with M* OO modules (Moose, Moo, Mo). With it, you structure your app
as one or more modules, which get instantiated with the commandline
arguments as attributes. Arguments are parsed using Getopt::Long::Descriptive,
and both long and short help messages as well as complete manual pages
are automatically generated. An app can be a single command with options,
or have sub-commands (like git). Sub-commands can be defined as modules
(with options of their own) or as simple coderefs.
Searches upward from the calling module for a directory t with
a lib directory inside it, and adds it to the module search path.
Looks upward up to 5 directories. This is intended to be used in
test modules either directly in t or in a subdirectory to
find their included testing libraries located in t/lib.
Add missing DEPENDS
Upstream changes:
0.9721 2021-04-18
- fix BitMatrix and AdjacencyMatrix problems - thanks @dod38fr for report
0.9720 2021-03-25
- better fix - no mutate inputs
0.9719 2021-03-25
- fix all_paths infinite loop on cycle - thanks @tobez for report
0.9718 2021-03-13
- remove doc of deleted average_degree method - thanks @lindleyw for report
0.9717 2021-01-27
- bulk APIs for UnionFind
- add unionfind config option for util/grand.pl (benchmark-ish script)
- GRAPH_ALLOW_RECURSION env var to turn off recursion protection
- "Light" edge-map now uses bit-vectors -> smaller storage
- directed hypergraphs
- fix same_biconnected_components logic when given >2 vertices
0.9716 2021-01-01
- use Set::Object
- {neighbours,successors,predecessors,reachable}_by_radius
0.9715 2020-12-31
- fix AdjacencyMap::Light attributes so delete when path deleted
- fix as_hashes undirected edges: now both directions
- subgraph_by_radius take multiple vertices
0.9714 2020-12-25
- remove "omni*" - hypergraphs are simply directed or undirected
- as_hashes works with undirected hypergraphs
- add_edge with != 2 vertices only for undirected hypergraph
- any_edge
- delete_*_attributes_by_id (and deleting last attribute) now don't destroy that entity
- AdjacencyMap::Light can have attributes, so no slowdown if use (eg APSP)
0.9713 2020-12-19
- fix edges_at on self-edges in scalar context
- fix refvertexed_stringified predicate
- remove "hypervertices": a collection of n vertices is a hyperedge
- AdjacencyMap.get_paths_by_ids
- transitive_closure et al no longer re-bless objects to Graph
- AdjacencyMap.get_ids_by_paths
- no more uniqedged configurability
- BitMatrix transpose option
- Transitive closure records path successor, not predecessor. Method name and docs updated.
0.9712 2020-12-05
- bug-fix: set_edge_attribute_by_id add_edge_by_id if not exist
- connected_component_by_index behaves same with/without unionfind
- AdjacencyMatrix handle multiedged
- reduce redundant sorting for _UNORD, fix AdjacencyMap::Vertex with ID 0
- AdjacencyMap.stringify
- allow constructor args to override "prototype" object
- fix docs for TransitiveClosure to correctly say path_vertices default true
- AdjacencyMatrix now always creates adjacency matrix (clue in name)
- remove compat02 features
- drop untested scalar-context Traversal.postorder mutation behaviour
- much more lazy-loading of modules
- set_vertex_attribute_by_id now works on hypervertexed
- internal AdjacencyMap uses array not hash for mapping index to path
- successors/predecessors/rename_path work right with multivertex
- AdjacencyMap array -> stable vertices ordering, TCM performance benefit
- TransitiveClosure etc handle multiedged
- all_paths ignore self-loops
0.9711 2020-11-27
- ingest handle multivertexed, multiedged right
0.9710 2020-11-27
- all_paths method
- as_hashes handle multivertexed, multiedged right
0.9709 2020-11-22
- add path_count option to TransitiveClosure
- get_{edge,vertex}_attributes undef if no such entity, in list context
- as_hashes method
- ingest method
0.9708 2020-11-06
- update metadata for Test::More version dep
- stringify hypervertices right
- add rename_vertex, rename_vertices
0.9707 2020-10-31
- can't use Safe, ergo Storable, on 5.8
0.9706 2020-10-20
- metadata list test-deps as not runtime
0.9705 2020-10-20
- document clustering_coefficient return empty list on no vertices - RT#114094
- apply lazy-load patch from Stephen Loyd - RT#123236
- depend on Heap 0.80 instead of local fork. Heap::Elem is really an interface not superclass.
- fix uninitialised value warning in SP_Dijkstra - RT#118539
- fix complement losing vertices on unconnected graphs - RT#115366
- fix average_path_length when two vertices given - RT#120611
- switch to GitHub issues rather than RT
- fix all_successors with non-truthy names - fix#5
- add Graph::Matrix->stringify to help debug
- fix APSP_Floyd_Warshall logic error when subpaths totalled 0 - fix#3
- typo fix - thanks @jkeenan (#6)
- Added "use strict; use warnings;", etc - thanks @manwar (closes#2)
v101
----
- `BinaryenSetFunctionTable` and `module.setFunctionTable` have been removed
in favor of `BinaryenAddTable` and `module.addTable` respectively.
- `BinaryenIsFunctionTableImported` is removed.
- A new type `BinaryenElementSegmentRef` has been added to the C API with
new apis in both C & JS:
- `BinaryenAddActiveElementSegment`
- `BinaryenAddPassiveElementSegment`
- `BinaryenRemoveElementSegment`
- `BinaryenGetElementSegment`
- `BinaryenGetElementSegmentByIndex`
- `BinaryenElementSegmentGetName`
- `BinaryenElementSegmentSetName`
- `BinaryenElementSegmentGetTable`
- `BinaryenElementSegmentSetTable`
- `BinayenElementSegmentIsPassive`
- `module.addActiveElementSegment`
- `module.addPassiveElementSegment`
- `module.removeElementSegment`
- `module.getElementSegment`
- `module.getElementSegmentByIndex`
- `module.getTableSegments`
- `module.getNumElementSegments`
- `binaryen.getElementSegmentInfo`
- `BinaryenAddTable` and `module.addTable` no longer take offset and function
names.
- `BinaryenGetNumFunctionTableSegments` is replaced with
`BinaryenGetNumElementSegments`.
- `BinaryenGetFunctionTableSegmentOffset` is replaced with
`BinaryenElementSegmentGetOffset`.
- `BinaryenGetFunctionTableSegmentLength` is replaced with
`BinaryenElementSegmentGetLength`.
- `BinaryenGetFunctionTableSegmentData` is replaced with
`BinaryenElementSegmentGetData`.
- Boolean values in the C API now should use `bool` instead of `int`.
- Experimental SIMD instructions have been removed and the names and opcodes of
the standard instructions have been updated to match the final spec.
0.20.21:
* Add basic support for a GcsObjectStore that stores
pack files in gcs.
* In porcelain.push, default to local active branch.
* Support fetching symrefs.
* Add aarch64 wheel building.
v1.6.4
For test dependencies, when indicating Python 3, use ``>=3``
instead of ``>3`` to satisfy
`python-poetry/poetry#3862 <https://github.com/python-poetry/poetry/issues/3862>`_.
v1.6.3
Restore universal wheel.
v1.6.2
Packaging refresh.
What's New in astroid 2.5.3?
============================
* Takes into account the fact that subscript inferring for a ClassDef may involve __class_getitem__ method
* Reworks the `collections` and `typing` brain so that `pylint`s acceptance tests are fine.
* Use ``inference_tip`` for ``typing.TypedDict`` brain.
* Fix mro for classes that inherit from typing.Generic
* Add inference tip for typing.Generic and typing.Annotated with ``__class_getitem__``
2.3 (2021-04-13)
================
* Do type hinting the right way.
2.2 (2021-04-13)
================
* ``public()`` and ``private()`` can't be correctly type annotated, so the
type hints on these two functions have been removed. The ``ModuleAware``
was also removed.
* Added a ``py.typed`` file to satisfy type checkers.
* Fixed a documentation cross-reference bug.
5.4.0 (2021-04-15)
==================
- Make the C implementation of the ``__providedBy__`` descriptor stop
ignoring all errors raised when accessing the instance's
``__provides__``. Now it behaves like the Python version and only
catches ``AttributeError``. The previous behaviour could lead to
crashing the interpreter in cases of recursion and errors.
- Update the ``repr()`` and ``str()`` of various objects to be shorter
and more informative. In many cases, the ``repr()`` is now something
that can be evaluated to produce an equal object. For example, what
was previously printed as ``<implementedBy builtins.list>`` is now
shown as ``classImplements(list, IMutableSequence, IIterable)``.
- Make ``Declaration.__add__`` (as in ``implementedBy(Cls) +
ISomething``) try harder to preserve a consistent resolution order
when the two arguments share overlapping pieces of the interface
inheritance hierarchy. Previously, the right hand side was always
put at the end of the resolution order, which could easily produce
invalid orders.
While the link fix did fix the case of openssl calling nss code,
the other way round still happens, e.g. in libreoffice (since fixed to
not use nss) and konqueror.
Bump PKGREVISION.
0.29.23 (2021-04-14)
====================
Bugs fixed
----------
* Some problems with Python 3.10 were resolved.
* An incorrect "optimisation" was removed that allowed changes to a keyword
dict to leak into keyword arguments passed into a function.
* Multiplied str constants could end up as bytes constants with language_level=2.
* ``PY_SSIZE_T_CLEAN`` does not get defined any more if it is already defined.
5.0.0 (2021-03-19)
==================
- Remove backwards compatibility imports that were emitting
deprecation warnings. This affects certain imports from
``zope.component.interfaces`` (which should be imported from
``zope.interface.interfaces``) as well as certain imports from
``zope.component.registery`` (import from
``zope.interface.registry``), and the entire
``zope.component.hookable`` module.
- Respect permission value for utility factory registrations
- Add support for Python 3.9
- Fix the ``<subscriber>`` ZCML directive to allow a missing
``provides=`` attribute when a ``factory=`` is given and the Python
object has been decorated with ``@implementer`` and implements a
single interface. This has been documented, but hasn't worked
before.
- Make ``PersistentAdapterRegistry`` use persistent objects
(``PersistentMapping`` and ``PersistentList``) for its internal data
structures instead of plain dicts and lists. This helps make it
scalable to larger registry sizes.
This requires zope.interface 5.3.0a1 or later.
New registries (and their primary users, ``PersistentComponents`` and
zope.site's ``LocalSiteManager``) take full advantage of this
automatically. For existing persistent registries to take advantage
of this, you must call their ``rebuild()`` method and commit the
transaction.
- Fix ``zope.interface.interface.provideInterface`` and the various
search and query methods to use the current site manager instead of
always using the global site manager. (``provideInterface`` is
called implicitly when registering components from ZCML.) The search
and query methods continue to return interfaces registered in base
site managers.
5.3.0 (2021-03-17)
==================
- Add support for Python 3.9.
- Fix `package init file missing` warning
- Make standard streams provide a `buffer` attribute on Python 3 when using
`--buffer` or testing under subunit.
5.3.0 (2020-03-21)
==================
- No changes from 5.3.0a1
5.3.0a1 (2021-03-18)
====================
- Improve the repr of ``zope.interface.Provides`` to remove ambiguity
about what is being provided. This is especially helpful diagnosing
IRO issues.
- Allow subclasses of ``BaseAdapterRegistry`` (including
``AdapterRegistry`` and ``VerifyingAdapterRegistry``) to have
control over the data structures. This allows persistent
implementations such as those based on ZODB to choose more scalable
options (e.g., BTrees instead of dicts). See `issue 224
<https://github.com/zopefoundation/zope.interface/issues/224>`_.
- Fix a reference counting issue in ``BaseAdapterRegistry`` that could
lead to references to interfaces being kept around even when all
utilities/adapters/subscribers providing that interface have been
removed. This is mostly an issue for persistent implementations.
Note that this only corrects the issue moving forward, it does not
solve any already corrupted reference counts. See `issue 227
<https://github.com/zopefoundation/zope.interface/issues/227>`_.
- Add the method ``BaseAdapterRegistry.rebuild()``. This can be used
to fix the reference counting issue mentioned above, as well as to
update the data structures when custom data types have changed.
- Add the interface method ``IAdapterRegistry.subscribed()`` and
implementation ``BaseAdapterRegistry.subscribed()`` for querying
directly registered subscribers. See `issue 230
<https://github.com/zopefoundation/zope.interface/issues/230>`_.
- Add the maintenance method
``Components.rebuildUtilityRegistryFromLocalCache()``. Most users
will not need this, but it can be useful if the ``Components.utilities``
registry is suspected to be out of sync with the ``Components``
object itself (this might happen to persistent ``Components``
implementations in the face of bugs).
- Fix the ``Provides`` and ``ClassProvides`` descriptors to stop
allowing redundant interfaces (those already implemented by the
underlying class or meta class) to produce an inconsistent
resolution order. This is similar to the change in ``@implementer``
in 5.1.0, and resolves inconsistent resolution orders with
``zope.proxy`` and ``zope.location``. See `issue 207
<https://github.com/zopefoundation/zope.interface/issues/207>`_.
Overview of changes in GLib 2.68.1
==================================
* Fix a crash in `GKeyFile` when parsing a file which contains translations
using a `GKeyFile` instance which has loaded another file previously
* Pin GIO DLL in memory on Windows (!2016)
* Bugs fixed:
- g_key_file_load_from_file segfaults on "Key[*]="like lines
- !1997 Backport !1996 Include glibconfig.h to get the G_OS_UNIX token to glib-2-68
- !2016 GIO W32: Pin gio DLL
- !2021 Backport MR !2016 (GIO W32: Pin gio DLL) into glib-2-68
- !2022 Few fixes and notes for building on Visual Studio 2012 and earlier
- !2034 Backport MR !2032 (gkeyfile: Drop a redundant check) into glib-2-68
- !2035 Backport !2026 Split g_test_log() messages that contain multiple lines to glib-2-68
* Translation updates:
- English (United Kingdom)
0.14.1
Known Open Issues (Blocker or Critical)
THRIFT-3877 - C++: library don't work with HTTP (csharp server, cpp client; need cross test enhancement)
THRIFT-5098 - Deprecated: "The high level Network interface is no longer supported. Please use Network.Socket." and other Haskell issues
THRIFT-5245 - NPE when the value of map's key is null
Deprecated Languages
THRIFT-5347 - Deprecate Haskell bindings
Build Process
THRIFT-5334 - version of thrift-maven-plugin is not sync with the main project
Delphi
THRIFT-5350 - 0.14.0 fails to build on non-x86
Go
THRIFT-5353 - Namespace from type is ignored in generated code
Python
THRIFT-5352 - Python: IDL exceptions with no fields can't be instantiated
Rust
THRIFT-5299 - rs implementation compact protocol seq_id should not use zigzag encoding.
