Skip to content

Voice-to-text dictation app with local (Nvidia Parakeet/Whisper) and cloud models (BYOK). Privacy-first and available cross-platform.

fix(settings): Clamp the LLM request timeout input on blur

Gabriel Stein70320880b7ec
diff --git a/src/components/SettingsPage.tsx b/src/components/SettingsPage.tsx
index e67ccb353..f96038b8f 100644
--- a/src/components/SettingsPage.tsx
+++ b/src/components/SettingsPage.tsx
@@ -113,6 +113,7 @@ import { syncService } from "../services/SyncService.js";
import {
LLM_REQUEST_TIMEOUT_SECONDS_MIN,
LLM_REQUEST_TIMEOUT_SECONDS_MAX,
+ resolveLlmRequestTimeoutSeconds,
} from "../helpers/llmRequestTimeout.js";
import { formatBytes } from "../utils/formatBytes";
import {
@@ -731,6 +732,13 @@ function LlmsTabs({
step={5}
value={llmRequestTimeoutSeconds}
onChange={(e) => setLlmRequestTimeoutSeconds(Number(e.target.value))}
+ // Clamp on blur so the stored value always matches the effective
+ // one (point of use clamps to 10-600s; a cleared field stores 0).
+ onBlur={() =>
+ setLlmRequestTimeoutSeconds(
+ resolveLlmRequestTimeoutSeconds(llmRequestTimeoutSeconds)
+ )
+ }
className="w-20"
/>
</SettingsRow>