v0.4.0

Changes

CLI is the default mode again. Bare ollama-profiler now errors asking for a model name (or --tui), matching pre-v0.3.0 behavior. This is friendlier for agent and script usage where silently dropping into a full-screen TUI was surprising.

• --tui flag reintroduced; launches the interactive TUI.
• --dry-run now requires --tui.
• Help text, examples, and install hints updated.

Kept from v0.3.0

• --tui rejects --json / --html / --png / --export (these were silently ignored before).
• Duplicate export path detection via filepath.Abs + filepath.Clean so ./out.json and out.json collide as expected.
• publish.yml runs go test before binaries/wheels build; contents: write scoped to the release job; if-no-files-found: error on artifact uploads.

Migration

If you relied on ollama-profiler alone to launch the TUI, add --tui:

ollama-profiler --tui

v0.3.1

Bug Fixes

• Fix --think silently accepting invalid values — Typos like --think medum now error instead of silently disabling thinking. Valid values: true, low, medium, high.
• Fix --num-predict 0 sending Ollama default instead of unlimited — 0 now correctly sends -1 to Ollama for unlimited generation.
• Reject negative --num-predict and --seed — Previously silently fell back to Ollama defaults; now returns a clear validation error.
• Reject --dry-run in CLI mode — Previously accepted but silently ignored; now errors with guidance to use TUI mode.
• Normalize --think to lowercase — --think=HIGH now works correctly instead of falling through to disabled.
• Distinguish "not installed" vs "not running" in connection errors — Shows platform-specific install instructions when Ollama binary isn't found, scoped to localhost URLs only.
• Surface Ollama error messages on non-200 responses — Error messages from Ollama (e.g., "model does not support thinking") are now shown instead of generic HTTP status errors.
• Guard BuildSchedule against empty model slice — Prevents a panic in balanced multi-round mode with no models.

Improvements

• IPv4-only test server helper for better CI portability
• 16 new regression tests covering flag validation, request serialization, and error paths

Full Changelog: v0.3.0...v0.3.1

v0.3.0

What's Changed

• Fix CI workflow assertions by @piercecohen1 in #2
• Add LICENSE, fix install.sh archive format, align Makefile targets by @piercecohen1 in #3
• Make TUI the default mode, harden CI and export flags by @piercecohen1 in #4

Full Changelog: v0.2.0...v0.3.0

v0.2.0

What's New

CLI Export Flags

• --json FILE — Export raw results with metadata to JSON
• --html FILE — Export self-contained HTML report with summary table and bar charts
• --png FILE — Export charts as a retina-quality PNG (generated natively, no browser needed)
• --export DIR — Export full bundle (JSON + HTML + PNG) to a directory

--export is mutually exclusive with --json / --html / --png (validated at startup).

Shared Export Package

• Extracted export logic from TUI into internal/export/ — now shared by both CLI and TUI modes
• TUI simplified by ~400 lines
• 23 integration tests covering all export formats, edge cases, and helpers

Other

• Bundle() returns (string, error) so callers get the directory path back
• NaN handling: zero-duration rates stored as NaN to avoid poisoning averages; JSON serializes them as null

Full Changelog: v0.1.0...v0.2.0

v0.1.0

Full Changelog: https://github.com/piercecohen1/ollama-profiler/commits/v0.1.0