Packages changed:
  Mesa (26.1.3 -> 26.1.4)
  Mesa-drivers (26.1.3 -> 26.1.4)
  blog (2.42 -> 2.45)
  cJSON
  dhcp (4.4.2.P1 -> 4.4.3.P1)
  geoclue2 (2.8.1 -> 2.8.2)
  gjs (1.88.0.g4 -> 1.88.1)
  glycin-loaders (2.1.1 -> 2.1.5)
  gnome-keyring
  gnome-shell (50.2 -> 50.3)
  gtk-vnc
  hwdata (0.408 -> 0.409)
  json-c (0.18 -> 0.19)
  kquickimageeditor6 (0.6.2 -> 0.6.2.1)
  libadwaita (1.9.1 -> 1.9.2)
  libglycin (2.1.1 -> 2.1.5)
  libglycin-gtk4 (2.1.1 -> 2.1.5)
  libisoburn (1.5.6 -> 1.5.8)
  libisofs (1.5.6.pl01 -> 1.5.8.pl01)
  libpsl (0.21.5 -> 0.22.0)
  libstorage-ng (4.5.336 -> 4.5.339)
  libwacom (2.18.0 -> 2.19.0)
  lilv (0.26.4 -> 0.28.0)
  mutter (50.2 -> 50.3)
  openSUSE-release (20260703 -> 20260707)
  openblas_openmp (0.3.30 -> 0.3.33)
  openblas_pthreads (0.3.30 -> 0.3.33)
  pam
  pam-full-src
  pinentry (1.3.2 -> 1.3.3)
  pinentry-gui (1.3.2 -> 1.3.3)
  python-greenlet
  python-service_identity (24.2.0 -> 26.1.0)
  rpm
  setxkbmap (1.3.4 -> 1.3.5)
  soundtouch (2.3.3 -> 2.4.1)
  sso-mib (0.9.0 -> 0.10.0)
  taglib (2.2.1 -> 2.3)
  upower (1.91.2 -> 1.91.3)
  vim (9.2.0725 -> 9.2.0780)

=== Details ===

==== Mesa ====
Version update (26.1.3 -> 26.1.4)
Subpackages: Mesa-libEGL1 Mesa-libGL1 libgbm1

- Update to 26.1.4 bugfix release
  - -> https://docs.mesa3d.org/relnotes/26.1.4

==== Mesa-drivers ====
Version update (26.1.3 -> 26.1.4)
Subpackages: Mesa-dri Mesa-libva Mesa-vulkan-device-select libvulkan_lvp

- Update to 26.1.4 bugfix release
  - -> https://docs.mesa3d.org/relnotes/26.1.4

==== blog ====
Version update (2.42 -> 2.45)
Subpackages: libblogger2

- Update to version 2.45
  Fix early boot LUKS prompt on s390x and resolve memory corruptions
  This update addresses several critical issues in the early boot phase,
  especially on s390x architectures, and hardens the daemon's architecture:
  - Initrd/Dracut: Properly include `systemd-ask-password-blog.path` in
    sysinit.target.wants during the initrd phase to ensure LUKS password
    prompts are captured early in the zipl phase.
  - Systemd Units: Drop restrictive `ConditionKernelCommandLine=!plymouth.enable=0`
    to prevent the agent from being disabled by standard mainframe boot parameters.
  - Memory Corruption: Fix a critical heap corruption by renaming the conflicting
    `alignof()` macro to `ALIGNED_SIZEOF()` and zeroing allocated memory for
    `struct request` (memset) to prevent reading uninitialized pointers #5.
  - Kernel Command Line: Improve `parse_cmdline()` to support boolean parameters
    without an explicit value (e.g., `blog.silent` defaults to `blog.silent=1`).
  - New Features: Introduce `blog.silent` to suppress console I/O and `blog.coldboot`
    to explicitly trigger the early coldstart password query.
  - Lifecycle: Switch `KillMode=none` to `KillMode=mixed` to comply with modern
    systemd process lifecycle management.
  Fixes: gh#bitstreamout/showconsole#5
  Fixes: gh#bitstreamout/showconsole#6
  Fixes: bsc#1264176
- Update to version 2.44
  Harden blog and use shims units to handle plymouth
  Disable coldstart requests via epoll
  Avoid handling fd twice in epoll loop
- Correct latest changes of 2.43
- Update to version 2.43
  Make sure that if blog is running it is identified as plymouth Latest
  also add conflicts to systemd-ask-password-console units in the
  systemd-ask-password-blog units. This should avoid conflicting password
  agents asking the same question in s390x.

==== cJSON ====

- Add baselibs.conf (bsc#1270256)

==== dhcp ====
Version update (4.4.2.P1 -> 4.4.3.P1)
Subpackages: dhcp-relay dhcp-server

- Update dhcrelay script and service.
  Due to some leftover calls to sysvinit tools the script no longer
  worked. Now the script is a lot leaner and only implements the
  sysconfig arg parsing and makes it more of a native systemd
  service.
  [bsc#1269228]
- Update to version 4.4.3-P1:
  * Corrected reference count leak in leasequery packets (CVE-2022-2928).
  * Corrected memory leak in FQDN option unpacking (CVE-2022-2929).
  * Added OMAPI timed function calls.
  * Updated BIND libraries to version 9.11.36.
  * Added support for DHCPv4 option v6-only-preferred (RFC 8925).
- Drop obsolete patches already included in version 4.4.3-P1:
  * dhcp-CVE-2022-2928.patch
  * dhcp-CVE-2022-2929.patch
- Refresh patches:
  * 0009-dhcp-4.2.6-close-on-exec.patch
  * 0015-Expose-next-server-DHCPv4-option-to-dhclient-script.patch
  * 0016-infiniband-support.patch

==== geoclue2 ====
Version update (2.8.1 -> 2.8.2)
Subpackages: system-user-srvGeoClue typelib-1_0-Geoclue-2_0

- Update to version 2.8.2
  New features:
  * Translations for Georgian an Occitan and instructions for
    translators.
  Fixes:
  * Building translations now works.
  * Fix double free in on_desktop_id_set() and
    on_client_proxy_ready().
  * Fix some gobject-linter rules in the correctness category.
  * Document the command to register network NMEA sources to avahi
- handle language files

==== gjs ====
Version update (1.88.0.g4 -> 1.88.1)
Subpackages: libgjs0 typelib-1_0-GjsPrivate-1_0

- Update to version 1.88.1:
  + Closed bugs and merge requests:
  - Breaking change introduced in 1.87.1
  - maint: Check for NULL before dereferencing error_out pointer

==== glycin-loaders ====
Version update (2.1.1 -> 2.1.5)

- Update to version 2.1.5:
  + Fixed: image-rs/jpeg: Support RGB for editing. There are
    actually images that don't use ycbcr.

==== gnome-keyring ====
Subpackages: gnome-keyring-32bit gnome-keyring-lang gnome-keyring-pam gnome-keyring-pam-32bit libgck-modules-gnome-keyring

- Add gnome-keyring-Drop-X-GNOME-Autostart-Phase.patch
  gnome-session-service is not responsible anymore for starting
  session services and complains about the presence of X-GNOME-Autostart-Phase.
  Removing this will allow whatever else is responsible (systemd,
  systemd-xdg-autostart-generator or whatever) to act.

==== gnome-shell ====
Version update (50.2 -> 50.3)
Subpackages: gnome-extensions gnome-shell-calendar gnome-shell-lang

- Update to version 50.3:
  + Prevent workspaces from being removed mid-animation
  + Announce search result description to screen readers
  + Improve animation when expanding app grid label
  + Fix spacing in lock screen media notifications
  + Plugged leak
  + Misc. bug fixes and cleanups
  + Updated translations.

==== gtk-vnc ====
Subpackages: gtk-vnc-lang libgtk-vnc-2_0-0 libgvnc-1_0-0 libgvncpulse-1_0-0

- bsc#1268264 - virt-manager crash when connecting to instance
  using VNC
  009-fix-error-handling-in-main-thread-context.patch
- bsc#1258497 - virt-manager segfault error 4 in libgvnc-1.0.so.0.0.1
  Renamed 009-let-GLib-manage-the-lifecycle-of-VncDisplay-GObject.patch to
  010-let-GLib-manage-the-lifecycle-of-VncDisplay-GObject.patch

==== hwdata ====
Version update (0.408 -> 0.409)

- Update to version 0.409:
  * Update pci, usb and vendor ids

==== json-c ====
Version update (0.18 -> 0.19)

- Add no-xxd.patch and bring back the full testsuite.
- Name the -devel and -doc subpackages based on the SRPM and
  nothing else.
- Drop xxd BuildRequires to break an OBS build cycle; exclude the
  test_json_parse_cli test (its only consumer) from %check.
- Update to 0.19:
  New features
  - -----------
  * Add support for Commodore Amiga and MorphOS
  * Allow builds with CMake 4 - sync minimum version across all
    CMakeLists.txt files
  * Update openssl command to work for newer versions of openssl.
  * Add support for building json-c with meson
  * Support MSVC in packages that use GNU Autoconf.
  * Add VERSIONINFO to libjson-c.dll
  * Make json_tokener_free(NULL) a no-op, to simplify cleanup paths.
  * Explicitly handle NaN values when converting to int
  * Set errno=RANGE in json_object_get_int/int64/uint64() when the
    source value can't be represented in the target type.
  * Make json_parse a bit more useful by adding -u (validate UTF8)
    and -P (specify arbitrary tokener parse flags), and read from
    stdin if no filename is provided.
  * Fix the apps/json_parse "-s" (strict) option so it actually does
    something, and default to non-strict.
  Significant changes and bug fixes
  - --------------------------------
  * Issue #867 - also disallow control characters in keys in
    JSON_TOKENER_STRICT mode
  * Issue #875: cast to unsigned char so bytes above 0x7f aren't
    interpreted as negative, which was causing the strict-mode
    control characters check to incorrectly trigger.
  * Issue #881: don't allow json_tokener_new_ex() with a depth < 1
  * Fix linkhash breaking -std=c89
  * Fixing Cmake build when using clang-cl, avoids errors about
    redefining existing symbols
  * Fix AIX build failure - Add CMake detection for getopt.h
  * Fix bug involving supplemental code points that look like high
    surrogates
  * Fix runtime issue with test_util_file.c in Windows (add O_BINARY)
  * Fix macro WIN32_LEAN_AND_MEAN redefined
  * Issue #914: Fix Memory usage regression due to newlocale() on
    macOS
  * Issue #916: Fix OOM via large array index in json_pointer_set
  * Issue #923: Avoid stack recursion in json_object_put()
  * Issue #927: CVE-2026-9146 - update json_object_iterator
    documentation
  * Issue #929: CVE-2026-11322 - fix information disclosure bug in
    apps/json_parse (not installed by default)
  * Issue #930: fix locale-dependent strtod in json_object_get_double
  * Issue #931: deep copy values in json_patch copy op to avoid
    aliasing and cycles

==== kquickimageeditor6 ====
Version update (0.6.2 -> 0.6.2.1)
Subpackages: kquickimageeditor6-imports libKQuickImageEditor1

- Udpate to 0.6.2.1
  * Fix a crash caused by lack of ARGB32 support when built with
    OpenCV

==== libadwaita ====
Version update (1.9.1 -> 1.9.2)
Subpackages: libadwaita-1-0 libadwaita-lang typelib-1_0-Adw-1

- Update to version 1.9.2:
  + Build:
  - Include prebuilt styles again
  - Resolve locale directory at runtime on Windows
  + AdwApplication: Speed up initial style loading
  + AdwStyleManager:
  - Speed up initial loading
  - Check GTK version in runtime instead of build time
  + AdwSwipeTracker: Ignore mostly orthogonal touchpad scrolls
  + AdwViewSwitcherSidebar: Fix a leak
  + Stylesheet: Fix a copypaste error
  + Updated translations.

==== libglycin ====
Version update (2.1.1 -> 2.1.5)

- Update to version 2.1.5:
  + Fixed: image-rs/jpeg: Support RGB for editing. There are
    actually images that don't use ycbcr.

==== libglycin-gtk4 ====
Version update (2.1.1 -> 2.1.5)

- Update to version 2.1.5:
  + Fixed: image-rs/jpeg: Support RGB for editing. There are
    actually images that don't use ycbcr.

==== libisoburn ====
Version update (1.5.6 -> 1.5.8)
Subpackages: libisoburn1 xorriso

- update to 1.5.8:
  * Bug fix: -boot_image and -append_partition were not perceived as image
    manipulation which makes production of an ISO image worthwhile.
    Thanks Cameron Seader.
  * Bug fix: -outdev holding an ISO filesystem could get attached wrong start LBA
  * Bug fix: Command -load "volid" did not work with constant search text
  * Bug fix: Command -truncate_overwritable on ISO image without MD5 caused
    double free of memory
  * Bug fix: -boot_image "any" "replay" failed after the legacy BIOS boot image
    file was replaced by -map. Thanks Brian C. Lane.
  * Bug fix: -boot_image system_area=/dev/zero preserved system area of
    loaded ISO
  * Bug fix: Size assessment of ISO images smaller than 32 KiB yielded
    random values
  * Bug fix: -extract_boot_images reports "No such file or directory" when
    complaining over an existing target file
  * Bug fix: Reproducibility of ISO images depended on "undefined" qsort(3)
    behavior with comparer function result 0.
    Report and fix proposal by Henrik Lindström.
  * Bug fix: ATA and SATA drives on modern Linux refused to perform -rollback
    with -scsi_dev_family "sg" because libburn registered them as ATA
    without SCSI adress tuple
  * Bug fix: Symbolic links with xattr could cause unusable ISO filesystems.
    Thanks Paul von Behren.
  * Bug fix: When reading multiple slashes in symbolic link targets from an ISO
    they were condensed to a single one if not marked by the
    SL ROOT-flag. Thanks Jonas 'Sortie' Termansen.
  * Bug fix: Empty symbolic link target components which represent add-on
    slashes were written with ROOT flag set.
    Thanks Jonas 'Sortie' Termansen.
  * Bug fix: The last trailing slash in a symbolic link target was not written.
    Thanks Jonas 'Sortie' Termansen.
  * Bug fix: The .. directory record exposed Rock Ridge properties of its
    grand grand parent. Caused by commit 4d0063f7 in 2009 which fixed
    the block address pointer of the .. record.
  * New command -volid_for and -as mkisofs option -volid_for.
  * New command -perform_opc
  * New -as cdrecord options --perform_opc and --perform_opc_growisofs

==== libisofs ====
Version update (1.5.6.pl01 -> 1.5.8.pl01)

- update to 1.5.8.pl01:
  * Bug fix: iso_image_set_ignore_aclea() inverted the setting
    for reading the XFS-style project id from disk files
  * Silenced error messages about ioctl(FS_IOC_FSGETXATTR)
  * Bug fix: File path of imported BIOS boot image was forgotten
    when it gets overwritten by a file of the same name.
  * Bug fix: Reproducibility of ISO images depended on
    "undefined" qsort(3) behavior with comparer function result 0
  * Bug fix: Empty AAIP value spoiled decoding of all AAIP
    attributes
  * Bug fix: Symbolic links with xattr could cause unusable ISO
    filesystems.
  * Bug fix: When reading multiple slashes in symbolic link
    targets they were
  * condensed to a single one if not marked by the SL ROOT-flag
  * Bug fix: Empty symolic link target components which represent
    add-on slashes were written with ROOT flag set
  * Bug fix: The last trailing slash in a symbolic link target
    was not written
  * Bug fix: The .. directory record exposed Rock Ridge
    properties of its
  * grand grand parent. Caused by commit 4d0063f7 in 2009 which
    fixed the block address pointer of the .. record.
  * Bug fix: The characters '.' and ';' were wrongly disallowed
    with PVD file names
  * Improved size estimation and property assessment of El Torito
    hidden boot images
  * Enabled up to 8 appended GPT partitions and improved their
    number mapping
  * New API calls iso_local_get_lfa_flags(),
    iso_local_set_lfa_flags(),
  * iso_node_get_lfa_flags(), iso_node_set_lfa_flags(),
  * iso_util_decode_lfa_flags(), iso_util_encode_lfa_flags(),
  * iso_util_get_lfa_masks(), iso_util_get_effective_lfa_mask()
  * New API calls iso_node_get_projid(), iso_node_set_projid(),
  * iso_local_get_projid(), iso_local_set_projid()
  * New API call iso_write_opts_set_gpt_with_gaps
  * New API call iso_local_create_dev()
  * New API call iso_write_opts_set_rrip_tf_long() and date check
    for year 2150
  * limit to produce 17-byte timestamps
  * New API call iso_write_opts_set_rrip_tf_y1900()
  * New API calls iso_image_set_volume_id_v2(),
    iso_image_get_volume_id_v2(),
  * iso_image_fs_get_volume_id_v2()
  * New API call iso_write_opts_set_relaxed_nonvol_atts()

==== libpsl ====
Version update (0.21.5 -> 0.22.0)

- Update to version 0.22.0:
  * Can now be built from the release tarball without Python
  * Drop an external dependency by implementing the IPv4/IPv6
    address checks internally
  * Avoid using alloca() and improve general portability
  * Add support for native ICU on Windows and libcucore on Apple
    systems
  * Improve the meson and autotools builds and the installed
    libpsl.pc pkg-config file
  * Documentation improvements
  * No soname change (still libpsl.so.5)

==== libstorage-ng ====
Version update (4.5.336 -> 4.5.339)
Subpackages: libstorage-ng-lang libstorage-ng-ruby libstorage-ng1

- merge gh#openSUSE/libstorage-ng#1087
- improved logging
- 4.5.339
- merge gh#openSUSE/libstorage-ng#1086
- updated tests
- 4.5.338
- merge gh#openSUSE/libstorage-ng#1085
- provide functions for mkfs and tune options using vector of
  strings
- use string_view to reduce number of functions
- 4.5.337

==== libwacom ====
Version update (2.18.0 -> 2.19.0)
Subpackages: libwacom-data libwacom9

- update to 2.19.0:
  * libwacom_list_styli_from_database() provides a WacomStylus
    list of all known styli
  * libwacom_database_ref() and libwacom_database_unref() to
    simplify callers having
  * libwacom_get_width_mm() and libwacom_get_height_mm() for
    the tablet dimensions in mm. The .tablet files switched to
    mm too.
  * New generic 3 button + eraser tool, assigned to some Huion
    devices by default
  * New devices:
  - Gaomon M5 V2
  - HP Envy x360 15-fh0xxx
  - Huion Inspiroy Frego S (L310), Kamvas Pro 27 (GT2701), H610
    Pro V2
  - Lenovo ThinkBook 14s Yoga ITL
  - Wacom HID 53FD
  - Xencelabs Pen Display 16

==== lilv ====
Version update (0.26.4 -> 0.28.0)

- update to 0.28.0:
  * Add early assertions for non-null parameters in public API
  * Add lilv_state_get_bundle_path()
  * Always create state files with O_CLOEXEC on glibc >= 2.7
  * Clarify relative symbolic link creation when saving state
  * Fix build with dynmanifest support
  * Fix crash when loading plugin classes on Windows
  * Fix potential file loss when saving state with no link
    directory
  * Fix potential iterator leaks and resulting log message flood
  * Fix replacing links when saving over state
  * Fix test build when no C++ compiler is available
  * Improve error handling when writing and deleting state

==== mutter ====
Version update (50.2 -> 50.3)
Subpackages: mutter-lang

- Update to version 50.3:
  + Fix erroneous stack order changes on workspace switches
  + De-prioritize focusing desktop windows
  + Ensure preferred_scale event is sent to clients
  + Fix incorrect accumulated damage in screencasts
  + Fix triggering OSK for text_input_v3 version 1 clients
  + Fixed crashes
  + Misc. bug fixes and cleanups
- Drop patches applied upstream:
  + ba64e055b5159a3acd5b4e4bc3d381066b26c62c.patch
  + 5096.patch

==== openSUSE-release ====
Version update (20260703 -> 20260707)
Subpackages: openSUSE-release-appliance-custom openSUSE-release-dvd

- automatically generated by openSUSE-release-tools/pkglistgen

==== openblas_openmp ====
Version update (0.3.30 -> 0.3.33)

- disable tests for i586 as sblat tests fail
- Added compatlib packages (compatlibopenblas_serial0, compatlibopenblas_openmp0,
  compatlibopenblas_pthreads0) that provide update-alternatives support for
  backward compatibility (bsc#1268086) (bsc#1268086)
- Compatlib packages are automatically installed via Supplements when upgrading
  from OpenBLAS >= 0.3.30 to maintain existing update-alternatives configurations
- The main library packages no longer require update-alternatives, making them
  suitable for transactional systems. Users who need update-alternatives support
  can explicitly install the compatlib packages
- Fixed pkgconfig and cmake symlinks to point directly to flavor-specific
  directories (e.g., /usr/lib64/openblas-openmp/) instead of through
  update-alternatives, ensuring consistent library resolution
- Updated README.SUSE to document the new packaging structure and compatlib usage
- Update to version 0.3.33
  general:
  * fixed incorrect cast in SBGEMM test case causing spurious failures
  * fixed invalid memory access in converted C CBLAS tests
  * made BIGNUMA setting automatic when cores exceed 256
  * imported recent Reference-LAPACK 3.13.0 updates including:
  - implemented LARF1F and ORM2R functions
  - changed loop order in GETC2 for better performance
  - added NaN checks for input matrix A in GEEV
  - fixed support for jobu/v in LAPACKE_GESVDQ_WORK
  - fixed workspace size calculation in TGSEN
  - optimized looping in LACPY/LASCL/LANTR with fat matrix and UPLO=L
  arm64:
  * worked around serious GCC15 miscompilation of DDOT kernel affecting most non-SVE targets
  * fixed accuracy issue in GEMV kernel for Neoverse V1 and SVE targets
  * fixed broken STRMM and SSYMM in DYNAMIC_ARCH builds on non-SME hardware
  * added optimized SHGEMM kernel for Neoverse N2
  * added autodetection of Cortex A75/A76 and Neoverse V3
  loongarch64:
  * fixed build failure from missing half-precision float type support
  * fixed long-standing bug in asserting 64bit capability in c_check helper
  x86_64:
  * added workaround for LLVM miscompilation of AVX512 GEMM kernels
  * fixed CMake compilation with NVIDIA HPC compiler
- Update to version 0.3.32
  general:
  * fixed LAPACK GESV reimplementation to compute LU factorization even when NRHS is zero
  * improved error message when compile-time buffer allocation exceeded
  * fixed non-serialized access problem in parallelized SYRK by concurrent callers
  * fixed ABI mismatch in internal CDOT/ZDOT version used by C fallback LAPACK
  * improved f_check script for detecting Fortran compiler with embedded dashes in paths
  * fixed memory access issues in utests detected by Address Sanitizer
  * fixed Makefile errors when only subset of precision types selected
  * fixed OpenMP compiler option for Intel Ifx compiler
  arm64:
  * fixed potential miscompilation of SDOT/DDOT/DSDOT kernels
  * improved BGEMM and SBGEMM kernel performance for Neoverse V2
  * added optimized SSUM and DSUM kernels for Neoverse N1
  * added preliminary Neoverse V3 support as NEOVERSEV2
  * added cpu autodetection for Cortex A725 and X925
  * fixed compilation of GEMM kernels for VORTEXM4/ARMV9SME without multithreading
  loongarch64:
  * fixed POTRF returning wrong results on LA464 from incorrect parameter setting
  power:
  * fixed compilation problems from missing half-precision float support
  * fixed potential POWER10 DGEMV kernel miscompilation by limiting optimization level
  * fixed SCAL issue on PPCG4/PPC970 running Linux
  riscv64:
  * optimized CROT/ZROT kernel for vector length 128 in non-unit stride path
  * improved SBGEMM/SHGEMM and related type conversion helper functions
  * fixed probing for BFLOAT16 support in DYNAMIC_ARCH cpu detection
  x86_64:
  * fixed potential gcc 15.x miscompilation of AVX512 SGEMM kernel for small sizes
  * fixed SROT and DROT kernels for Haswell to have consistent FMA rounding
  * added automatic detection of Intel Arrow Lake H/U, Panther Lake, and Jasper Lake
  * added detection of Intel Emerald Rapids and upcoming models
  * updated cache size translation table in cpu model autodetection
  * improved cpu detection fallback to include Nehalem as non-AVX option
  * fixed Makefile build issue with clang and SkylakeX SGEMM kernel
- Update to version 0.3.31
  general:
  * reverted matrix partitioning optimization from 0.3.30 that could cause race conditions in GEMM
  * added bfloat16 extensions BGEMM and BGEMV
  * added BLAS interface for GEMM_BATCH extensions
  * added GEMM_BATCH_STRIDED extensions and CBLAS interface
  * added basic infrastructure for half-precision float (FP16) format using SH prefix
  * reimplemented LAPACK SLAED3/DLAED3 using multithreading, improving SSYEVD/DSYEVD performance
  * limited memory allocation retries to avoid infinite hanging on low-memory systems
  * fixed thread lockup with python 3.9 or older and numpy
  * introduced problem size thresholds for multithreading in STRMV/DTRMV, CHER/ZHER functions, and SGER/DGER
  * improved autodetection of Fortran compiler
  * fixed INTERFACE64=1 option passing to flang-new compiler
  * fixed potential deadlock after fork() in multithreaded code
  * fixed CMake builds on FreeBSD and from Cygwin
  * fixed cross-compilation to RISCV targets via CMake
  * fixed installation of cblas.h when NO_CBLAS=0
  * fixed missing user-defined pre/postfixes on functions in lapack.h and lapacke.h
  arm64:
  * building BFLOAT16 kernels now enabled by default
  * improved overall GEMM, SYMM and HEMM performance on A64FX
  * improved SDOT/DDOT performance on A64FX
  * improved multithreading performance of SDOT/DDOT on A64FX via throttling table
  * improved SGER/DGER performance on A64FX and NEOVERSEV1
  * improved multithreading GEMM performance on A64FX and NEOVERSEV1
  * improved GEMV kernel for SVE-capable targets
  * improved multithreading SGEMM on NEOVERSEV1 and V2
    ... changelog too long, skipping 31 lines ...
  * added support for building with CMake

==== openblas_pthreads ====
Version update (0.3.30 -> 0.3.33)

- disable tests for i586 as sblat tests fail
- Added compatlib packages (compatlibopenblas_serial0, compatlibopenblas_openmp0,
  compatlibopenblas_pthreads0) that provide update-alternatives support for
  backward compatibility (bsc#1268086) (bsc#1268086)
- Compatlib packages are automatically installed via Supplements when upgrading
  from OpenBLAS >= 0.3.30 to maintain existing update-alternatives configurations
- The main library packages no longer require update-alternatives, making them
  suitable for transactional systems. Users who need update-alternatives support
  can explicitly install the compatlib packages
- Fixed pkgconfig and cmake symlinks to point directly to flavor-specific
  directories (e.g., /usr/lib64/openblas-openmp/) instead of through
  update-alternatives, ensuring consistent library resolution
- Updated README.SUSE to document the new packaging structure and compatlib usage
- Update to version 0.3.33
  general:
  * fixed incorrect cast in SBGEMM test case causing spurious failures
  * fixed invalid memory access in converted C CBLAS tests
  * made BIGNUMA setting automatic when cores exceed 256
  * imported recent Reference-LAPACK 3.13.0 updates including:
  - implemented LARF1F and ORM2R functions
  - changed loop order in GETC2 for better performance
  - added NaN checks for input matrix A in GEEV
  - fixed support for jobu/v in LAPACKE_GESVDQ_WORK
  - fixed workspace size calculation in TGSEN
  - optimized looping in LACPY/LASCL/LANTR with fat matrix and UPLO=L
  arm64:
  * worked around serious GCC15 miscompilation of DDOT kernel affecting most non-SVE targets
  * fixed accuracy issue in GEMV kernel for Neoverse V1 and SVE targets
  * fixed broken STRMM and SSYMM in DYNAMIC_ARCH builds on non-SME hardware
  * added optimized SHGEMM kernel for Neoverse N2
  * added autodetection of Cortex A75/A76 and Neoverse V3
  loongarch64:
  * fixed build failure from missing half-precision float type support
  * fixed long-standing bug in asserting 64bit capability in c_check helper
  x86_64:
  * added workaround for LLVM miscompilation of AVX512 GEMM kernels
  * fixed CMake compilation with NVIDIA HPC compiler
- Update to version 0.3.32
  general:
  * fixed LAPACK GESV reimplementation to compute LU factorization even when NRHS is zero
  * improved error message when compile-time buffer allocation exceeded
  * fixed non-serialized access problem in parallelized SYRK by concurrent callers
  * fixed ABI mismatch in internal CDOT/ZDOT version used by C fallback LAPACK
  * improved f_check script for detecting Fortran compiler with embedded dashes in paths
  * fixed memory access issues in utests detected by Address Sanitizer
  * fixed Makefile errors when only subset of precision types selected
  * fixed OpenMP compiler option for Intel Ifx compiler
  arm64:
  * fixed potential miscompilation of SDOT/DDOT/DSDOT kernels
  * improved BGEMM and SBGEMM kernel performance for Neoverse V2
  * added optimized SSUM and DSUM kernels for Neoverse N1
  * added preliminary Neoverse V3 support as NEOVERSEV2
  * added cpu autodetection for Cortex A725 and X925
  * fixed compilation of GEMM kernels for VORTEXM4/ARMV9SME without multithreading
  loongarch64:
  * fixed POTRF returning wrong results on LA464 from incorrect parameter setting
  power:
  * fixed compilation problems from missing half-precision float support
  * fixed potential POWER10 DGEMV kernel miscompilation by limiting optimization level
  * fixed SCAL issue on PPCG4/PPC970 running Linux
  riscv64:
  * optimized CROT/ZROT kernel for vector length 128 in non-unit stride path
  * improved SBGEMM/SHGEMM and related type conversion helper functions
  * fixed probing for BFLOAT16 support in DYNAMIC_ARCH cpu detection
  x86_64:
  * fixed potential gcc 15.x miscompilation of AVX512 SGEMM kernel for small sizes
  * fixed SROT and DROT kernels for Haswell to have consistent FMA rounding
  * added automatic detection of Intel Arrow Lake H/U, Panther Lake, and Jasper Lake
  * added detection of Intel Emerald Rapids and upcoming models
  * updated cache size translation table in cpu model autodetection
  * improved cpu detection fallback to include Nehalem as non-AVX option
  * fixed Makefile build issue with clang and SkylakeX SGEMM kernel
- Update to version 0.3.31
  general:
  * reverted matrix partitioning optimization from 0.3.30 that could cause race conditions in GEMM
  * added bfloat16 extensions BGEMM and BGEMV
  * added BLAS interface for GEMM_BATCH extensions
  * added GEMM_BATCH_STRIDED extensions and CBLAS interface
  * added basic infrastructure for half-precision float (FP16) format using SH prefix
  * reimplemented LAPACK SLAED3/DLAED3 using multithreading, improving SSYEVD/DSYEVD performance
  * limited memory allocation retries to avoid infinite hanging on low-memory systems
  * fixed thread lockup with python 3.9 or older and numpy
  * introduced problem size thresholds for multithreading in STRMV/DTRMV, CHER/ZHER functions, and SGER/DGER
  * improved autodetection of Fortran compiler
  * fixed INTERFACE64=1 option passing to flang-new compiler
  * fixed potential deadlock after fork() in multithreaded code
  * fixed CMake builds on FreeBSD and from Cygwin
  * fixed cross-compilation to RISCV targets via CMake
  * fixed installation of cblas.h when NO_CBLAS=0
  * fixed missing user-defined pre/postfixes on functions in lapack.h and lapacke.h
  arm64:
  * building BFLOAT16 kernels now enabled by default
  * improved overall GEMM, SYMM and HEMM performance on A64FX
  * improved SDOT/DDOT performance on A64FX
  * improved multithreading performance of SDOT/DDOT on A64FX via throttling table
  * improved SGER/DGER performance on A64FX and NEOVERSEV1
  * improved multithreading GEMM performance on A64FX and NEOVERSEV1
  * improved GEMV kernel for SVE-capable targets
  * improved multithreading SGEMM on NEOVERSEV1 and V2
    ... changelog too long, skipping 31 lines ...
  * added support for building with CMake

==== pam ====
Subpackages: pam-32bit

- Fix password comparison timing leak from pam_userdb
  [CVE-2026-54411], [bsc#1268290]
  * pam_userdb-fix-password-comparison-timing-leak.patch

==== pam-full-src ====
Subpackages: pam-extra pam-manpages

- Fix password comparison timing leak from pam_userdb
  [CVE-2026-54411], [bsc#1268290]
  * pam_userdb-fix-password-comparison-timing-leak.patch

==== pinentry ====
Version update (1.3.2 -> 1.3.3)

- Update to 1.3.3:
  * gtk,curses: Add --focus=ok option to CONFIRM command. [T8315]
  * tty: Discard input on timeout. [rP4c72137885]
  * Various build fixes.

==== pinentry-gui ====
Version update (1.3.2 -> 1.3.3)
Subpackages: pinentry-gnome3 pinentry-gtk2 pinentry-qt6

- Update to 1.3.3:
  * gtk,curses: Add --focus=ok option to CONFIRM command. [T8315]
  * tty: Discard input on timeout. [rP4c72137885]
  * Various build fixes.

==== python-greenlet ====


==== python-service_identity ====
Version update (24.2.0 -> 26.1.0)

- update to 26.1.0:
  * Python 3.14 and 3.15 are now officially supported.
  * *service-identity* now uses *cryptography*'s Rust-based ASN.1
    decoder and doesn't depend on *pyasn1* and *pyasn1-modules*
    anymore.
  * As a result, the oldest supported pyOpenSSL backend
    combination is now *pyOpenSSL* 26.1.0 with *cryptography*
    47.0.0.
  * Verifying a single-label hostname (e.g. `localhost`) against
    a wildcard certificate pattern now raises `VerificationError`
    cleanly instead of crashing with an opaque `ValueError`.

==== rpm ====
Subpackages: librpmbuild10

- Add Requires: (rpm-plugin-selinux if selinux-policy)
- harden ndb code [bsc#1269584] [CVE-2026-44605]
  * new patch: ndbharden.diff
- split all plugins into subpackages
  * this allows for an easy way to get rid of a plugin, it's also
    what other distributions do
- make the imaevmsign plugin build in a multibuild flavor
  * updated patch: imaevmsignplugin.diff
- rpm2archive: use size 0 for hardlinked files as bnew versions of
  gnu tar reject non-zero sizes [bsc#1269150]
  * updated patch: rpm2archive.diff
- backport fix for add_sysuser macro [bsc#1269571]
  * updated patch: macrosin.diff
- backport rpmuncompress security fix [bsc#1268747] [CVE-2026-44604]
  * new patch: rpmuncompress.diff
- switch from rpmpgp_legacy to libpgpr
  * multiple bug fixes, support for v5 and v6 signatures
  * updated patch: pgpreleasemtime.diff
- turn on imaevm file signature support and move the imaevm code
  that needs the libimaevm library into a plugin. Put this
  plugin into a new "rpm-imaevmsign" subpackage. [jsc#PED-7246]
  * new patch: imaevmsignplugin.diff

==== setxkbmap ====
Version update (1.3.4 -> 1.3.5)

- update to 1.3.5:
  * Man page: update history from XFree86 4.3 to 3.2
  * setxkbmap references appear as far back as 3.2 in the XFree86
    Changelog.
  * manual page: Rewrite in semantic markup
  * Rewrite the reference manual in BSD-style mdoc(7) semantic
    markup, providing greatly enhanced search on all supported manual
    compilers.
  * Improve man page formatting
  * More closely follow common style as described on
  * meson: Add option to build with meson

==== soundtouch ====
Version update (2.3.3 -> 2.4.1)

- update to 2.4.1:
  * Fix buffer overflow when processing >16 channels
  * fix: fix unused variables in DLMain
  * fix: fix warning for unused peaklevel in release mode
- Update to 2.4.0:
  * Set CMake minimum version to 3.5 to avoid deprecation warning
  * Increase max nr. of channels from 16 to 32
  * Don't use `pow()` when using integer precision samples
  * Replace `-Ofast` that's being deprecated in some compilers, by
    `-O3 -ffast-math`
- Refresh disable-ffast-math.patch

==== sso-mib ====
Version update (0.9.0 -> 0.10.0)

- Import version 0.10.0
  This release adds an async API for all functions that communicate
  via DBus. The API is modeled similar to gdbus. Manpages for tools
  are also now provided.

==== taglib ====
Version update (2.2.1 -> 2.3)
Subpackages: libtag2 libtag_c2

- update to 2.3:
  * MP4: Support for chapters (Nero and QuickTime).
  * WAV: Support for BEXT and iXML chunks.
  * FLAC: Support for BEXT and iXML application blocks.
  * Opus: New audio property `outputGain()`.
  * Speed up Matroska reading by using seek head for element
    lookup.
  * Speed up Matroska writing by offering multiple write style
    modes.
  * More tolerant handling of files with oversized RIFF chunks,
    zero size ID3v2
  * frames and Matroska chapters without edition.
  * Avoid wrong content-based detection as MPEG files.
  * Fix bitrate calculations for MPEG ADTS and MP4 ESDS.
  * Fix data race with multi-threaded use of `MP4::ItemFactory`.
  * Fix unbounded recursion in EBML/Matroska `MasterElement` and
    MP4 atoms.
  * Limit number of MP4 atoms at top level.
  * Fix writing too many offsets when updating MP4 stco/co64
    atoms.
  * Fix k bounds in Shorten Rice-Golomb coding.

==== upower ====
Version update (1.91.2 -> 1.91.3)
Subpackages: libupower-glib3 typelib-1_0-UpowerGlib-1_0 upower-lang

- Update to version 1.91.3:
  + Feature: up-device-battery: Prefer "Standard" over "Fast"
    charging
  + Fix: Resolve potential leaks
  + Fix: Potential out-of-bound access
  + Fix: Improve access control
  + Fix: Remove unused codes
  + Fix: Fix for Asus Battery Charge Threshold Detection
- Drop 25303ba52771ee514b70fb1a5318a8313889ac31.patch: Added
  upstream.

==== vim ====
Version update (9.2.0725 -> 9.2.0780)
Subpackages: vim-data vim-data-common xxd

- update to 9.2.0780
  patch 9.2.0780: Memory leak in evalvars.c on alloc failure
  patch 9.2.0779: Memory leak in type_name_func() on alloc failure
  patch 9.2.0778: Memory Leak in compile_dict() on alloc failure
  patch 9.2.0777: Memory leak in add_defer() on alloc failure
  patch 9.2.0776: Memory leak in sign_getlist() on alloc failure
  patch 9.2.0775: Memory Leak in highlight_get_info() on alloc failure
  patch 9.2.0774: Memory leak in f_getscriptinfo() on alloc failure
  patch 9.2.0773: Memory leak in evalfunc.c on alloc failure
  patch 9.2.0772: Vim9: Null dereference inside alloc_type()
  patch 9.2.0771: dict_add_list() has inconsistent ownership on failure
  patch 9.2.0770: dict_add_dict() has inconsistent ownership on failure
  translation(zh_CN): Add Chinese man pages
  patch 9.2.0769: conversion to utf-16be using iconv is inconsistent
  patch 9.2.0768: legacy/vim9cmd modifiers are not exclusive
  patch 9.2.0767: legacy/vim9cmd modifiers do not set script version for options values
  patch 9.2.0766: quick_tab entries for empty letters point to the wrong index
  patch 9.2.0765: popup: opacity popup over a terminal is not cleared when moved
  patch 9.2.0764: Compiler warning about unused function
  translation(sr): Update Serbian translation
  patch 9.2.0763: tests: style issue in test_plugin_netrw
  runtime(xml): Update xml syntax file
  runtime(doc): regenerate help tags
  patch 9.2.0762: duplicated sub-option name check in :set completion
  runtime(screen): Bring the syntax up to version 5
  runtime(typst): Improve ftplugin, and syntax file
  patch 9.2.0761: runtime(netrw): Unix: unable to open '\' file
  patch 9.2.0760: Compiler warning for using potentially uninitialized var
  patch 9.2.0759: Some code for 'autocompletedelay' is no longer needed
  runtime(autoload/dist): gx may use xdg-open on macOS
  patch 9.2.0758: pum: No opacity when background not set for Popup menu group
  patch 9.2.0757: tests: test_popupwin fails with zsh because of the prompt
  patch 9.2.0756: Session with multiple tabpages sets 'winminheight' to 0
  patch 9.2.0755: 'autocomplete' behaves inconsistently when recording
  patch 9.2.0754: repeated completion length lookup in search_for_exact_line
  patch 9.2.0753: GTK GUI deferred redraw skipped on 'lazyredraw'
  runtime: add missing fnameescape()/shellescape() in a few runtime files
  patch 9.2.0752: GTK4: drag-and-drop does not support HTML
  patch 9.2.0751: GTK3 GUI is slow under Wayland
  translation(ru): Updated message files
  patch 9.2.0750: completion: 'autocompletedelay' deferral leaks state
  patch 9.2.0749: 'autocompletedelay' interferes with i_CTRL-K
  patch 9.2.0748: 'autocompletedelay' interferes with CTRL-G U
  runtime(netrw): Use consistent forward slashes
  patch 9.2.0747: cscope: connection leak when growing the array fails
  patch 9.2.0746: NULL pointer dereference in gui_photon
  patch 9.2.0745: Crash with truncated spellfile
  patch 9.2.0744: popup_atcursor() closes immediately on white space
  runtime(odin): Update indent script, add indent tests
  patch 9.2.0743: string macros silently accept a size of the wrong type
  runtime(vim): Fix heredoc triggering misidentifcation of Vim9 script
  patch 9.2.0742: filetype: SSH keys and related filetypes not recognized
  runtime(css): add more missing CSS properties
  patch 9.2.0741: complete_check() does not return TRUE for mapped input
  patch 9.2.0740: GTK4: scrollbar wrongly displayed
  patch 9.2.0739: completion: 'autocompletedelay' blocks the main loop and drops autocommands
  runtime: guard recommended style settings consistently
  patch 9.2.0738: ml_recover() may write beyond block buffer
  CI: Bump actions/checkout in the github-actions group across 1 directory
  patch 9.2.0737: tests: comment test can be improved
  patch 9.2.0736: potential command execution in PHP omni-completion
  patch 9.2.0735: [security]: arbitrary Ex command execution during C omni-completion
  runtime(doc): Tweak documentation style and typo
  patch 9.2.0734: function pointer passed to STRNCMP() instead of a length
  runtime(cabal): Update compiler, ftplugin, syntax, add indent script
  patch 9.2.0733: GTK3: GUI slow on X11 since dropping the alpha channel
  patch 9.2.0732: session: terminal restored using absolute columns/rows
  patch 9.2.0731: GTK4 GUI scrollbar size not updated when restoring a session
  patch 9.2.0730: GTK4 GUI tabline is not updated
  patch 9.2.0729: % skips parens on continued quoted lines
  runtime(doc): Add installer updates to version9.txt
  patch 9.2.0728: filetype: supertux info pattern is relative to current dir
  patch 9.2.0727: popup images not rendered correctly when unfocused
  patch 9.2.0726: filetype detect missing from completion