Upgrade to release 1.2.0:
- Add i8, u64, u32 and u16 native integer types
- Drop support for Python 3.8 and 3.9, officially support 3.14 and
3.15
- Permit building with flit core 4
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 5.7.2:
- Proper routing of an event called * in the catch-all handler
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 6.1.3:
Security fixes:
- CVE-2026-108147: Potential memory exhaustion in MultiPartParser
via base64-encoded file upload
- CVE-2026-111211: Potential log injection via unescaped geometry
form values
- CVE-2026-111362: Server-side request forgery via crs member of
GeoJSON geometrie
- CVE-2026-113593: Potential exposure of private data via empty
and missing Vary headers
- CVE-2026-113733: Password hash disclosure to “view only” admin
users with custom user models
- CVE-2026-114582: Server-side request forgery via
SpatialReference string input
Bug fixes:
- Fixed a bug in Django 6.1 where accessing a deferred field with
the FETCH_PEERS fetch mode overwrote values that peer instances
had already modified locally or loaded via refresh_from_db() with
an explicit fields argument. The fix also addresses a KeyError
when a peer's row had been deleted
- Fixed a regression in Django 6.1 where the YAML serializer
crashed if passed the sort_keys option
- Fixed a regression in Django 6.1 where a Value expression with a
None value and an output field other than JSONField crashed on
SQLite when used in a top-level JSONField in lookup. Using such an
expression in an IN lookup is likely unintended, since SQL NULL
cannot match any value.
- Fixed a regression in Django 6.1 where legend_tag() with a
custom attrs could add the required_css_class more than once.
- Added compatibility for oracledb 26.0.0+
- Fixed a crash in makemessages on Windows when processing files
with non-ASCII names on Python 3.15+, or with UTF-8 mode enabled
on older Python versions.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 7.17.1:
- Fix: a one-line class or function body immediately followed by
another statement (such as class Foo: ... on its own line) was
incorrectly reported as a missing branch when using the
sys.monitoring core. This is now fixed.
- Fix: the HTML report only rewrites a file's page when the file or
its data has changed, but each page also links to the previous
and next files. When the set of reported files changed, unchanged
pages kept their old links. For example, with skip_covered, a
file that became 100% covered had its page deleted, but the pages
next to it still linked to it. Now a page is also rewritten when
its neighbors change.
- Fix: --fail-under now uses at least the decimal precision of the
threshold for both the comparison and failure message. With
precision=0, a total of 0.4 now passes --fail-under=0.3 and fails
--fail-under=0.5 with the message "total of 0.4 is less than
fail-under=0.5". Previously, the comparison rounded the total to
zero and the message could claim "1 is less than 0".
- Fix: the HTML report escaped file names twice, so a name with &,
<, or quotes in it was shown with entities like &. Now they
are properly escaped only once.
- Fix: --source=/ measured no files. A file system root can't be a
source directory, so now it is an error.
- In an abundance of caution, prevent forged SQLite journal files
from deleting files they shouldn't.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 4.4.1:
- Size the relay test's payload from the pipe capacity
- Publish releases from GitHub Actions with PyPI provenance
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 26.11.1:
- Security: parse the HTTP headers of the WebSocket opening
handshake strictly (RFC 9112, RFC 9110). The header block was
split with str.splitlines(), which also splits on VT, FF, FS, GS,
RS, NEL and on a bare CR or LF, so a peer could make autobahn see
header lines that a front proxy treats as part of a header value,
for example an X-Forwarded-For used with trustXForwardedFor.
Lines are now separated by CRLF only. ⚠ behaviour change — a
header block with a bare CR or LF, obs-fold, whitespace before
the colon, an invalid header name, a header line without a colon
(was skipped), or a control character in a header value or the
status line is rejected: the server fails the handshake with HTTP
400, the client fails the handshake or the proxy CONNECT. A bare
LF is rejected deliberately, although RFC 9112 allows accepting
it as a line end. obs-text (0x80-0xFF) in header values is still
allowed, and header values are trimmed of SP and HTAB only. This
fixes advisory GHSA-2gqc-mwv3-66rp
- Security: raise the floors of the compress extra (and so all)
to brotli>=1.2.1 (CPython) and brotlicffi>=1.2.1.0 (PyPy), the
releases with the fix for CVE-2026-111497, an out-of-bounds write
in the brotli decoder that permits remote code execution.
permessage-brotli decodes data sent by the WebSocket peer; it is
not used unless an application offers or accepts it. The brotli
1.0.10 and 1.1.1 backports of the fix are not enough, since the
permessage-brotli receive path needs the decompression limit of
brotli 1.2. New tests check the declared floors and the installed
version
- Add tests for the HTTP header parsing of the WebSocket opening
handshake, on the server, the client and the explicit proxy
CONNECT, run on Twisted and asyncio
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 6.10.1:
- Recognize trace operations defined in view docstrings, instead
of silently dropping them
- Fix intersect OneOf and Equal choices in field2choices
- Drop support for marshmallow 3, which is EOL
- Drop support for Python 3.10
- Test against Python 3.15
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Tries to call sphinx-build during clean in docs directory and fails when
not installed on host nor included in HOSTTOOLS
Fixes do_configure failure when re-executed in existing WORKDIR
...
rm -f ./*.gcov
make[2]: Entering directory '/OE/tmp/work/x86_64-linux/ctags-native/6.2.20261011.0/sources/ctags-6.2.20261011.0/docs'
/bin/sh: 1: sphinx-build: not found
make[2]: Leaving directory '/OE/tmp/work/x86_64-linux/ctags-native/6.2.20261011.0/sources/ctags-6.2.20261011.0/docs'
make[2]: *** [Makefile:20: clean] Error 127
make[1]: *** [Makefile:8218: clean-local] Error 2
make[1]: Leaving directory '/OE/tmp/work/x86_64-linux/ctags-native/6.2.20261011.0/sources/ctags-6.2.20261011.0'
make: *** [Makefile:7175: clean-recursive] Error 1
ERROR: oe_runmake failed
WARNING: exit code 1 from a shell command.
The issue was introduced in p6.2.20260802.0 version with:
https://github.com/universal-ctags/ctags/commit/05a8061a581f626deada11da5300822bee20e458
which replaced "rm -rf $(BUILDDIR) $(EXTDIR)/__pycache__" in clean with
sphinx-build clean:
@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
Signed-off-by: Martin Jansa <martin.jansa@gmail.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
The test is parametrized over gateway types; it imports "remote" from
its tmp_path and then deletes that directory, but the module stays in
sys.modules. Every run after the first got the stale module back and
failed with "OSError: could not get source code". Drop the cached
module before importing it.
AI-Generated: Uses Claude Code
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
1.7.5 replaced py.process.ForkedFunc with multiprocessing, so swap the
python3-py runtime dependency for python3-multiprocessing. The sdist is
now named pytest_forked, the pypi class default, so drop
PYPI_PACKAGE_SDIST.
Backport the pending upstream fix for pytest 9.2, where TestReport
carries a NodeId object in _id that marshal cannot serialize; every
forked test crashed with "ValueError: unmarshallable object". This made
python3-starlette ptests report no results at all; all of them pass
now on qemux86-64.
AI-Generated: Uses Claude Code
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
The project dropped setup.py, switch to python_setuptools_build_meta.
The sdist is now named pytest_benchmark, which is the pypi class
default, so drop PYPI_PACKAGE_SDIST. License checksum change is due
to the copyright year update.
5.1.0 broke every test with pytest 9.2 and Python 3.14:
AttributeError: 'Function' object has no attribute '_nodeid'
PendingDeprecationWarning: FileType is deprecated.
which made ptests fail in python3-pint, python3-fastjsonschema,
python3-portion, python3-tblib, python3-construct and
python3-pydantic-core. These now all pass on qemux86-64.
AI-Generated: Uses Claude Code
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
4.2.9 needs hatch-fancy-pypi-readme as a build backend plugin, add
it to DEPENDS.
This also fixes ptest failures seen with 4.2.4, e.g.
test_post_open_failure_closes_fd[fstat] and the soft lock
lease/stale tests; all ptests pass now on qemux86-64.
AI-Generated: Uses Claude Code
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
This includes fix for CVE-2026-114550.
Added a patch for reproducibility issue from this release.
Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
This is a bug-fix release addressing a couple more build issues after the
conversion to meson. Not really necessary for meta-openembedded but
let's upgrade for consistency with upstream.
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
The generated softhsm2.conf only sets directories.tokendir and
objectstore.backend. SoftHSM warns about every other key it falls back to
the default for, and it logs through syslog(), so every process loading
the module during a build (rauc, extract-cert, ...) puts three
"Missing ... in configuration" lines into the build host's journal.
Set log.level, slots.removable and slots.mechanisms explicitly. The latter
two keep their defaults; log.level = ERROR keeps SoftHSM's INFO/WARNING
chatter out of the host journal while errors are still logged.
Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Johannes Schneider <johannes.schneider@leica-geosystems.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 2026.10:
- meson: Disable tests build by default
- Define attr dependency
- Test if merged overlay is usable after mounting
- Remove trusted xattrs from upperdir metacopies before copying
attributes
- Add Option To Use rsync for merge Operation
- Support merge metacopy file
Since version 2026.10, meson.build wraps the tests build in
"if get_option('tests')", so the tests are disabled by default and
do_configure:prepend is no longer needed. Keeping it also breaks
the build with:
ERROR: Expecting endif got eof.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 2026.09.22:
- generic: test bdev freeze count leak on nested thaw
- btrfs/255: add _cleanup function to kill the balance stress
process
- common/btrfs: exercise all algorithms in _btrfs_stress_defrag()
- common/btrfs: exercise all algorithms in
_btrfs_stress_remount_compress()
- btrfs: test that we don't create pointless compressed inline
extents
- overlay/081: add missing _require_scratch check
- btrfs/273: add missing _require_scratch check
- common/btrfs: require scratch before checking encoded reads
- f2fs/015,016: add missing _require_scratch checks
- generic/590,735,795: add missing _require_scratch check
- common: strip attr 2.6.0 --restore safety warnings
- generic/760: enable sub-block reads
- generic/263: enable sub-block reads
- generic/091: enable sub-block reads
- min_dio_alignment: add a -r option to query read alignment
- statx.h: update to latest kernel UAPI
- generic: test corruption detection using T10 protection
information
- generic: test I/O on devices with T10 protection information
- add a "pi" group
- common: add a _sysfs_block_integrity_path helper
- common: factor out a _bdev_disk_name helper
- common/scsi_debug: don't slow down I/O
- f2fs/009: fix race condition in orphan inode test
- common/rc: remove trailing space from _mread() default map_len
- common: fix spelling of SCRATCH_DEV_POOL
- f2fs/024: add missing execute permission
Upgrade to v2026.09.22, not v2026.10.15 because the latter's
commits are still on branch for-next and have not yet merged to
branch master.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
OpenSSL 4.1 deprecates ASN1_STRING_length(), which the OpenSSL 4 ASN1
opaque fix uses, and the build uses -Werror:
openssl_helper.c:1255:21: error: 'ASN1_STRING_length' is deprecated:
Since OpenSSL 4.1; use ASN1_STRING_get_length() [-Werror,-Wdeprecated-declarations]
Use ASN1_STRING_get_length() when building against OpenSSL 4.1 or newer.
AI-Generated: Uses Claude Code
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
tests/test_sequence_docs.py executes the code examples embedded in
docs/usage.rst, resolved relative to the tests directory. The file was
not installed, so all 26 tests in it failed with:
FileNotFoundError: [Errno 2] No such file or directory:
'/usr/lib/python3-cbor2/ptest/docs/usage.rst'
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Regenerate python3-cbor2-crates.inc to match the bundled Cargo.lock.
The previous crates list was stale (bigdecimal 0.4.10 vs 0.4.11 locked),
which made cargo metadata fail in do_compile with:
failed to select a version for the requirement `bigdecimal = "^0.4.10"`
(locked to 0.4.11)
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
2.112 2019-12-13
- Added the change log from any prior source that could be found. Formatted
the log to fit metacpan.org loose standards.
- Added .mailmap to correctly attribute all commits to Dianne Skoll when
using: git shortlog -s -e
- Added .gitattributes and .gitignore to assist with keeping the repo clean
- Change use vars qw() to our $whatever instead. 'our'
- Hide IO::WrapTie subclasses from PAUSE
- Rebuild Makefile.PL to contain all of the prerequisites. Make use of the
stub from Dist::Zilla
- Convert README to README.md
- Fix the documentation in the main module, IO::Stringy to better indicate
where to get info and how to use the module
- Fix the dist's META information to indicate the original author and license
- Add a LICENSE file
- Get rid of the set-version.pl file as it's no longer needed
- Add AppVeyor CI testing
- Add Travis CI testing
- update t/IO_InnerFile.t to use Test::More and a proper TEMP file (RT #103895)
Signed-off-by: Jason Schonberg <schonm@gmail.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Changelog:
- Fix panics in thin_dump when btree nodes point beyond the metadata device
- Fix overflows in IoEngine block range tracking and offset calculation,
which caused panics or reads at wrong offsets on damaged metadata in
thin_check, cache_check and other tools.
- Fix integer overflow in space map boundary checks
- Remove redundant leaf insertions in LeafWalker
Signed-off-by: Ankur Tyagi <ankur.tyagi85@gmail.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Dropped patch that is part of the upstream version.
Changelog:
- Support building with OpenSSL 4.0.0 and later.
- Bug fix; UTF-7 encoding mailbox conversion problems.
Signed-off-by: Ankur Tyagi <ankur.tyagi85@gmail.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>