v0.7.1

v0.7.1

v0.7.0

v0.7.0

v0.6.0

v0.6.0

v0.5.0

v0.5.0

v0.3.5

v0.3.5

v0.3.4

v0.3.4

v0.3.3

v0.3.3

v0.3.2

• Docs: screenshots of the GPUs (MIG) and Idle & waste views in the README.
• UI: wider default widths for the "VRAM held" and "Peak in window" columns so the headers are not truncated.

v0.3.1

• Fix: LICENSE copyright holder.
• Docs: README Features and Usage sections; repo homepage points at the npm page.

v0.3.0

First published release, as @tal-naeh/freelens-gpu-extension on npm.

• Sidebar group GPU with five views: Pods, GPUs (per device / MIG slice), Idle & waste, Allocation, Exporters.
• Per-device aggregation for dcgm-exporter (model, MIG profile, temperature) and per-process exporters.
• Rolling in-memory history drives the Idle & waste view ("idle for" from consecutive samples).
• Allocation joins nvidia.com/gpu capacity / allocatable with running pods' requests and measured busy devices.
• Exporters view exposes discovery probes, scrape latency, body size and errors.
• Shared DataGrid: click-to-sort, drag-to-resize (double-click resets, widths persisted), sticky header, hover text.
• Version badge in every page title.
• Fix: scrape uses the relative /api-kube proxy path; KubeJsonApi.forCluster is declared in the 1.10.3 typings but
  missing at runtime.
• GPU sections in the Pod and Node detail drawers.