53 lines
2.6 KiB
Markdown
53 lines
2.6 KiB
Markdown
# Legacy Hardware Validation
|
|
|
|
This report exists to close the remaining Phase 2 validation work against the target profile: older desktop hardware with limited RAM where Thisper must remain responsive and low-overhead.
|
|
|
|
## Validation Checklist
|
|
|
|
### Background Stability
|
|
|
|
- Scenario: Leave Thisper running in the tray/background for an extended period.
|
|
- Acceptance target: No crash, no hotkey loss, no visible UI freeze when restoring the window.
|
|
- Current evidence: User-reported runtime of more than 20 hours without API or stability failures.
|
|
- Status: `partially validated`
|
|
|
|
### Repeated Cross-App Rewrite
|
|
|
|
- Scenario: Use `Ctrl + Alt + R` repeatedly across plain-text targets such as Notepad and browser text boxes.
|
|
- Acceptance target: No destructive replacement on provider failure, no clipboard corruption after failures, stable repeated use without restart.
|
|
- Current evidence: Informal real-world usage is positive, but no structured pass is recorded yet.
|
|
- Status: `remaining`
|
|
|
|
### Long-Form Desktop Rewrite
|
|
|
|
- Scenario: Rewrite a multi-paragraph block through the main window and inspect diff/output toggles.
|
|
- Acceptance target: UI remains responsive during rewrite and the diff view reflects only actual edits.
|
|
- Current evidence: Main UI build path is working and has been used successfully, but no formal timing pass is logged yet.
|
|
- Status: `remaining`
|
|
|
|
### Successive Rewrites Without Restart
|
|
|
|
- Scenario: Perform multiple rewrites in sequence from both the main UI and the global shortcut flow.
|
|
- Acceptance target: No cumulative instability, no stale model state, no stuck loading state.
|
|
- Current evidence: Runtime observability is now present to support this pass, but the pass itself is still pending.
|
|
- Status: `remaining`
|
|
|
|
### Resource Footprint
|
|
|
|
- Scenario: Observe memory use and responsiveness on constrained hardware during idle, active rewrite, and tray-hidden states.
|
|
- Acceptance target: No runaway memory growth and acceptable perceived latency for short text.
|
|
- Current evidence: Build/runtime artifacts are large in development, but that is mostly `target/`; release footprint still needs explicit validation.
|
|
- Status: `remaining`
|
|
|
|
## What To Record During The Pass
|
|
|
|
- idle memory use
|
|
- active rewrite latency for short and medium inputs
|
|
- whether the tray/hotkey path remains responsive after long idle time
|
|
- whether clipboard restoration remains correct after both success and failure
|
|
- any target app classes that consistently fail or behave inconsistently
|
|
|
|
## Completion Rule
|
|
|
|
Phase 2 desktop validation is complete when each scenario above has an explicit observed result recorded here, including failures or caveats.
|