Script 'mail_helper' called by obssrc Hello community, here is the log from the commit of package gputaskman for openSUSE:Factory checked in at 2026-11-10 16:30:02 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Comparing /work/SRC/openSUSE:Factory/gputaskman (Old) and /work/SRC/openSUSE:Factory/.gputaskman.new.3759290 (New) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "gputaskman" Tue Nov 10 16:30:02 2026 rev:2 rq:1396526 version:0.3.5 Changes: -------- --- /work/SRC/openSUSE:Factory/gputaskman/gputaskman.changes 2026-11-09 16:08:18.609030658 +0100 +++ /work/SRC/openSUSE:Factory/.gputaskman.new.3759290/gputaskman.changes 2026-11-10 16:30:19.550555600 +0100 @@ -1,0 +2,17 @@ +Tue Nov 10 07:28:29 UTC 2026 - Martin Pluskal <[email protected]> + +- Update to version 0.3.5: + * Linux Intel GPU (i915/xe) support is now upstream, including + GT clock, throttle status, private iGPU memory accounting and + a CPU-package temperature fallback: drop + gputaskman-intel-i915-xe.patch. + * New system RAM/CPU panel with per-process RAM/CPU columns and + a 'gputaskman record' command for append-only JSON snapshots. + * More reliable identity: exact-LUID Windows ROOT display + adapter names, vLLM model IDs with parent API provenance, and + snapshot diffs rejecting different adapters and reused PIDs. + * Safer model inspection: deduplicated GGUF paths, null-safe + Ollama manifests, authenticated local llama-server probes. + * WSL2 NVML discovery in Nix, no NVML sweep on non-NVIDIA GPUs. + +------------------------------------------------------------------- Old: ---- gputaskman-0.3.4.tar.gz gputaskman-intel-i915-xe.patch New: ---- gputaskman-0.3.5.tar.gz ----------(Old B)---------- Old: a CPU-package temperature fallback: drop gputaskman-intel-i915-xe.patch. * New system RAM/CPU panel with per-process RAM/CPU columns and ----------(Old E)---------- ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Other differences: ------------------ ++++++ gputaskman.spec ++++++ --- /var/tmp/diff_new_pack.JqaGzZ/_old 2026-11-10 16:30:21.167623238 +0100 +++ /var/tmp/diff_new_pack.JqaGzZ/_new 2026-11-10 16:30:21.168623280 +0100 @@ -19,7 +19,7 @@ # Single-flavour application: build against the primary python only, tracking it as it moves. %define pythons %{primary_python} Name: gputaskman -Version: 0.3.4 +Version: 0.3.5 Release: 0 Summary: GPU task manager with per-process VRAM and LLM insights License: MIT @@ -28,9 +28,6 @@ # Upstream caps uv_build < 0.10.0, but Factory ships 0.13.x (upstream CI uses 0.13.6): keep only the floor. # PATCH-FIX-OPENSUSE gputaskman-uv-build-cap.patch -- relax uv_build upper bound for Factory Patch0: gputaskman-uv-build-cap.patch -# Intel i915/xe support on Linux, backported from upstream (merged 2026-11-09). -# PATCH-FIX-UPSTREAM gputaskman-intel-i915-xe.patch gh#shanselman/gputaskman#13 -- Intel i915/xe support backport -Patch1: gputaskman-intel-i915-xe.patch # Help-screen test renders markdown with linkify (textual test dep, not a runtime dep). BuildRequires: %{python_module linkify-it-py} BuildRequires: %{python_module nvidia-ml-py >= 13.615.71} ++++++ gputaskman-0.3.4.tar.gz -> gputaskman-0.3.5.tar.gz ++++++ ++++ 6098 lines of diff (skipped)
