@ondeinference / onde / commits / f29b4ea

fix(pulse): handle UQFF configs when reading the stream model id

The match over LoadedModelConfig in stream_message was exhaustive when it was written; the branch had no Uqff variant then. Merging development brought UQFF model support in, and the new arm was never updated, so cargo check failed with E0004 on every platform target. The blocking path's equivalent match already covers Uqff. This makes the streaming one agree with it. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Seto Elkahfi committed Sep 4, 2026 at 23:28 UTC f29b4ea2e0bb5247ff976c051a7033b00b9411b6
1 file changed +1
src/inference/engine.rs
+1
@@ -1204,6 +1204,7 @@ impl ChatEngine {
1204 let request = self::build_request(loaded, &user_message);
1205 let pulse_model_id = match &loaded.config {
1206 LoadedModelConfig::Gguf(c) => c.model_id.clone(),
1207 + LoadedModelConfig::Uqff(c) => c.model_id.clone(),
1208 #[cfg(target_os = "macos")]
1209 LoadedModelConfig::Isq(c) => c.model_id.clone(),
1210 };