Upgrade to release 1.3.3:
- restore toplevel re-exports for full 1.2 API compatibility
- docs: quote pip extras url in installation instructions
- ci: test python 3.9 in the matrix
- dbc: import BA_ BU_ attribute for unknown ecu
- j1939: fill unset bits with 1 on encode
- fix: fix canmatrix.utils import in xlsx
Fixes:
WARNING: python3-canmatrix-1.3.3-r0 do_check_backend: QA Issue:
inherits setuptools3 but has pyproject.toml specifying backend
setuptools.build_meta, use the correct class [pep517-backend]
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 2026.6:
- Cuba will not fall back 2026-11-01; stays on -04 through
2027-11-07.
- No more leap seconds, and leap second tables no longer expire.
- Improve 1965/1966 data for Brazil.
Changes to future timestamps:
- Cuba will not fall back to -05 on 2026-11-01 and instead plans to
maintain -04 through 2027-11-07.
Changes to past timestamps:
- In 1965 Brazil ended DST on 03-31 at 24:00 not 00:00, and in 1966
it ended DST on 03-01 at 01:00 not 00:00.
Changes to documentation and commentary:
- Documentation and commentary now reflect leap seconds' demise.
Changes to commentary:
- Discussion on timekeeping in northwestern Ontario has been updated
to reflect that provincial regulation has been issued to allow
these areas to stay aligned with America/Winnipeg and follow
Manitoba's adoption of year-round -05 after all.
- Discussion on timekeeping in Saskatchewan has been updated to
document its new Time Act, 2026 which comes into force on
2026-10-30. No change is expected.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 0.17.3:
Preview features:
- [suspicious] Add shared-clock-mock
Bug fixes:
- [pyupgrade] Avoid incorrect conversions to f-strings (UP032)
Performance:
- Buffer stdout in ruff format
- Specialize ASCII identifier lexing
- Speed up identifier and keyword lexing
Documentation:
- [flake8-bandit] Describe what the rule actually checks for
httpx (S113)
This work was sponsored by GOVCERT.LU.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 2026.6:
- IANA 2026f
- tz changes for Brazil
- Northwestern Ontario is expected to follow Manitoba
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 4.1.0:
- Drop support for EOL python 3.10
- Significant performance improvements
- Fix handling of Literal and Annotated for lazy imports
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 1.45.2:
- opentelemetry-api: limit traceparent header length to 512
characters in TraceContextTextMapPropagator during extraction
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 1.2.4:
- fallback: pass object_pairs_hook a list of pairs
- fallback: leave the object unread when a header type does not
match
- fallback: return bytes from Unpacker.read_bytes
- Run tracemalloc tests sequentially
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 4.68.0:
- [feaLib] Escape " and \ when writing names back out with asFea()
in FeatureNameStatement, STATNameStatement and
CVParametersNameStatement.
- [scaleUpem] Also scale the MATH lengths that are not
MathValueRecords (DelimitedSubFormulaMinHeight,
DisplayOperatorMinHeight, MinConnectorOverlap, etc.).
- [cffLib] Inherit the initial CFF2 vsindex from the Private DICT
when decompiling, specializing and instancing charstrings, and
fix programToCommands miscounting operands when several blend
operators precede the first stem or moveto.
- [instancer] Round CFF2 outlines and stem hints on absolute
positions rather than per relative operand, and keep CFF2 Private
DICT blend values at full precision until they are written back.
- [cffLib] Preserve CFF2 Private DICT arrays that use more than one
blend group.
- [feaLib] Raise FeatureLibError instead of looping forever on an
unclosed OS/2, BASE or STAT table block.
- [designspaceLib] When a variable font overrides some
public.fontInfo keys, merge them with the root-level font info
instead of replacing it.
- [instancer] Raise an error for FeatureVariations conditions of a
format other than the supported axis-range one, instead of
silently dropping them.
- [filesystem] Don't overwrite an unrelated sibling .zip when
writing a .ufoz or other non-.zip archive.
- [glyf] Fix exponential running time on composite glyphs that
reuse the same component at several nesting levels;
getCoordinates now raises TTLibError past 65535 composite points.
- [ttLib] Add support for the beyond-64k tables of the final ISO
Open Font Format 5th edition: uppercase companion tables (GLYF,
GVAR, HMTX, GSUB/GPOS extended layout, FeatureVariations 1.1, TTC
1.1/2.1, DMAP), a ttLib.beyond64k module and CLI, and support in
FontBuilder, merge, subset, instancer, qu2cu, removeOverlaps and
scaleUpem.
- [VARC] Keep addVarComponent() locations sparse, passing
RESET_UNSPECIFIED_AXES via a new keyword-only
resetUnspecifiedAxes pen argument.
- [woff] Require brotli >= 1.2.1 and brotlicffi >= 1.2.1.0, which
fix GHSA-9v8h-qqgm-4m35.
- Build Windows ARM64 wheels.
- [instancer] Fix AttributeError: ElidedFallbackNameID when
updating font names of a font with a version 1.0 STAT table; fall
back to the subfamily name.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 40.44.0:
- Fix: allow repeated digits in pt_BR CPF and RG generators
- Fix: keep en_CA SINs out of the 0 and 8 series
- Fix: use Luhn check digit for no_NO Postgiro (0000) BBANs
- Fix: en_GB VAT ID generation
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 4.15.1:
- Reject requests before the connection completes
- Address possible race condition when checking the last pong
response to a ping
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 1.9.0:
Behavior changes:
- A false value for a flag now overrides lower-priority sources
Previously, setting a flag (an arg that takes no value, such as
store_true, store_false, count or store_const) to false in an
environment variable or a higher-priority config file added nothing
to the command line, so a true for the same flag in a lower-priority
config file still turned it on. Now the false value counts as
setting the flag, and lower-priority config files no longer set it.
For example, VERBOSE=false now wins over verbose = true in a config
file. Args that take a value are unchanged.
- A bare -- in a list value is now an error
A list value from a config file or environment variable, such as
items = [a, --], used to put a bare -- on the command line, which
ended option parsing and turned the user's own command-line options
after it into positional args. It now fails with an error that
names the key.
Fixes:
- Options after -- on the command line no longer block other
sources. In prog -- --level, --level is a positional arg,
so environment variables and config files can still set --level.
- IniConfigParser.serialize() escapes % (#364), so values like
%Y-%m-%d read back unchanged, and a saved %(name)s is no longer
replaced by another key's value when the file is loaded.
- Saved config files keep store_false and store_const values.
write_config_file() now writes true or false according to
whether the flag was given, rather than the value the flag
stores.
- ConfigparserConfigFileParser reads files that only have a
[DEFAULT] section, including the files its own serialize()
writes.
- ConfigparserConfigFileParser round trip: values written by
serialize() read back the same in more cases (for example strings
that look like lists, and list elements containing # or ;), a
value that will be cut short by a # or ; comment now produces a
warning, and interpolation errors name only the key, not the
value, which may be a secret.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 45.1.2:
- Revert "When parsing DBC, make sure there are no unused VAL_ and
SIG_VALTYPE_ entries" due to regressions
- sym: refuse to write a message with more than one multiplexer
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 7.2.7:
- Various performance improvements
- Minor unit test improvements
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Commit 2330468507 aspired to move the
CMake files from /usr/share/glaze to /usr/lib/cmake/glaze. However, all
it did was remove the FILES:${PN}-dev assignment, which resulted in the
CMake files ending up in the target package.
Actually set the install path for the CMake files using
-Dglaze_INSTALL_CMAKEDIR:PATH. For some odd reason, it needs to be
realtive, or the installed files will not work as intended when
cross-compiling.
Also remove -DCMAKE_INSTALL_PREFIX and -DDESTDIR from EXTRA_OECMAKE as
they are already set correctly by cmake.bbclass.
Finally, restore -Dglaze_DEVELOPER_MODE=OFF. It does not seem to affect
the installed files, but generates less CMake rules.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
ICU 79.1 brings Unicode 18, which adds U+1DF95 LATIN SMALL LIGATURE
LONG S WITH DESCENDER S. It case-folds to U+00DF, so for the first time
the case closure of a BMP character contains a supplementary code point.
V8's gen-regexp-special-case build tool passes every member of that
closure to RegExpCaseFolding::Canonicalize(), which asserts a UTF-16
code unit, and aborts with
# Check failed: ch <= 0xffff.
Backport the V8 fix, adapted to the V8 13.6 bundled with Node.js 24.
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
The obj/bytecode.c rule in the wrk Makefile has no dependency on the
obj/ directory, so with parallel make luajit can run before obj/ has
been created and fails:
LUAJIT src/wrk.lua
luajit: .../obj/bytecode.c: No such file or directory
make: *** [Makefile:64: obj/bytecode.c] Error 1
wrk only supports in-tree builds (relative obj/ and src/ paths, B = S),
so create ${S}/obj via do_compile[dirs], keeping ${S} last so make
still runs from the source directory.
Signed-off-by: Vaibhav Jindal <vaibjind@qti.qualcomm.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
The usage example in gflags_completions.h hardcodes
/home/build/eng/bash/bash_completions.sh, a path from the original build
environment. It matches the HOME of the yoe build container, so
do_package_qa reports:
File /usr/include/gflags/gflags_completions.h in package gflags-dev
contains a reference to the build host HOME directory. [buildpaths]
Point the example at the installed gflags_completions.sh instead.
AI-Generated: Uses Claude Code
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 7.2.1:
- Fixed a regression in 7.2.0 where a subclass of MultiDictProxy or
CIMultiDictProxy that converted its argument in its own
__init__() raised TypeError before that __init__() ran; such a
subclass got an empty multidict until its __init__() set the
proxied one --.
Fix ptest-runner errors.
License-Update: Update LICENSE file. The license remains the same.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 4.13.5:
- Publish a GitHub Release with changelog notes, a PyPI link, and a
version comparison after each PyPI release.
Deselect failing tests from the run-ptest script for ptest-runner.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 0.9.7:
- ndb: basic support for XFRM interfaces
- ndb: basic support for object preload hooks
- ndb: basic support for prometheus metrics
- taskstats: fix name unregister_mask() -> deregister_mask()
- plan9: fids per session & inodes per FS
- ethtool: fix the off flags
- dhcp: support option 41 NIS servers
- wirouting: link features
- rtnl: support RTAX_CC_ALGO on route metrics
- rtnl: add missing NFT flags
Fix ptest-runner errors.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Upgrade to release 1.10.0:
Leaner URL-encoded form parsing:
Starlette 1.10.0 parses URL-encoded forms with less memory and CPU, no
longer drops the last field when a stream ends early, and lets you
compare FormData that holds several uploads under one key.
- Forms with many fields parse faster and use less memory (#3724).
FormParser builds each field in the parser callbacks instead of
queueing every event and replaying it. On CodSpeed, 1000 fields
went from 42.8 ms to 34.9 ms and from 63.5 KB to 54.5 KB, and the
max_fields limit case went from 6,880 B to 864 B.
- The last field is kept when the request stream ends without an
empty chunk (#3599). The parser is now finalized once the stream
is exhausted, so custom streams that never yield a final b"" no
longer lose their last field.
- FormData with repeated uploads can be compared (#3619). Comparing
two FormData objects with several UploadFile values under the
same key used to raise TypeError.
Behaviour changes:
- Multidict equality keeps the order of values under a repeated key
(#3619). Keys can still appear in any order, but
QueryParams("a=1&a=2") and QueryParams("a=2&a=1") are no longer
equal, matching what getlist() returns.
- starlette.formparsers.FormMessage is deprecated (#3724). The
parser no longer uses it, and FormParser.messages is gone.
Calling FormMessage(...) emits a StarletteDeprecationWarning.
Routing and URL generation:
- Path parameter values that look like a placeholder are kept as
they are (#3725). url_path_for("route", category="{id}", id="42")
on /{category}/{id} now returns /{id}/42 instead of /42/42.
Substitution also got faster, from 549 ns to 520 ns with two
parameters and from 1,078 ns to 779 ns with five.
- Route matching does less work per request (#3665, #3722). Match
members are compared by identity, cutting up to 4.9% of the
instructions per dispatch in the routing benchmarks.
- WSGIMiddleware strips root_path only at a path segment boundary
(#3718). A request to /application under a root_path of /app no
longer becomes /lication.
- FloatConvertor reports NaN and infinity correctly (#3716). NaN
was reported as a negative float.
Middleware, responses and config:
- Handled exceptions no longer keep requests alive (#3205).
Starlette clears the traceback after an exception handler runs,
so an exception instance reused across requests, such as a
module-level HTTPException, no longer pins per-request objects in
memory.
- Middlewares accept http.response.start without headers (#3721).
BaseHTTPMiddleware, GZipMiddleware and SessionMiddleware default
the omitted field to an empty list instead of raising KeyError.
- FileResponse ignores Range on methods other than GET and HEAD
(#3679). A POST endpoint that returns a file now sends the full
file, as RFC 9110 requires.
- Config strips one pair of matching outer quotes from env file
values (#3720). KEY='say "hi"' now reads as say "hi" instead of
say "hi.
Extend the ptest runtime dependencies to fix test cases from
tests/middleware/test_opentelemetry.py.
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Add recipe for OpenTelemetry Python API, release 1.45.1:
- Make AnyValue and AttributeValue explicit type aliases
- Fix quadratic time parsing of the tracestate header and discard
tracestate headers longer than 8192 characters
Signed-off-by: Leon Anavi <leon.anavi@konsulko.com>
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
PKGV = "4.3+git${GITPKGV}" counts the commits up to SRCREV, which is
always 1 with shallow clones, so successive SRCREVs only compare by hash
and gitpkgv warns:
WARNING: tvheadend: Shallow cloning enabled - gitpkgv.bbclass will not
generate sortable versions
Drop gitpkgv and let PKGV default to PV, to which the fetcher appends the
PR server AUTOINC and the revision (4.3+git0+cc60283368). The AUTOINC
starts below the commit count used so far, so bump PE to keep the upgrade
path from existing packages.
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
SRCREV is the 2.11.8 tag, so PKGV = "${GITPKGVTAG}" only evaluates to
2.11.8 with a full clone. With shallow clones it becomes 0.0-1-g<hash>,
a downgrade, and gitpkgv warns:
WARNING: freerdp: Shallow cloning enabled - gitpkgv.bbclass will not
generate sortable versions
Drop gitpkgv and let PKGV default to PV.
Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
Leon Anavi
Jason Schonberg
Markus Volk
Peter Kjellerstedt
Khem Raj
Vaibhav Jindal
Ankur Tyagi
Martin Siegumfeldt