speed up tests

it was nice to have ollama testing the format as an integration
in CI, but those are covered well by unit tests already and this
increased test times significantly

.github/workflows/ci.yml

@@ -76,27 +76,9 @@ jobs:
           restore-keys: |
             ${{ runner.os }}-cargo-build-
 
-      - name: Install Ollama
-        run: curl -fsSL https://ollama.com/install.sh | sh
-
-      - name: Start Ollama
-        run: |
-          # Run the background, in a way that survives to the next step
-          nohup ollama serve > ollama.log 2>&1 &
-          # Block using the ready endpoint
-          time curl --retry 5 --retry-connrefused --retry-delay 1 -sf http://localhost:11434
-
-      - name: Test Ollama Model
-        run: ollama run qwen2.5 hello || cat ollama.log
-
-      - name: Build Rust Project
-        run: cargo build
-
       - name: Run Tests
-        run: cargo test --verbose
-        env:
-          OLLAMA_MODEL: "qwen2.5"
-
+        run: cargo test
+
       ## TODO: Need to decide if we wanna error out on clippy warnings. It was not being used before. 
       # - name: Run Cargo Clippy (Lint)
       #   run: cargo clippy -- -D warnings

crates/goose/src/providers/google.rs

@@ -11,7 +11,7 @@ use serde_json::Value;
 use std::time::Duration;
 
 pub const GOOGLE_API_HOST: &str = "https://generativelanguage.googleapis.com";
-pub const GOOGLE_DEFAULT_MODEL: &str = "gemini-1.5-flash";
+pub const GOOGLE_DEFAULT_MODEL: &str = "gemini-2.0-flash-exp";
 
 #[derive(Debug, serde::Serialize)]
 pub struct GoogleProvider {

crates/goose/tests/providers.rs

@@ -75,12 +75,14 @@ lazy_static::lazy_static! {
 /// Generic test harness for any Provider implementation
 struct ProviderTester {
     provider: Box<dyn Provider + Send + Sync>,
+    name: String,
 }
 
 impl ProviderTester {
-    fn new<T: Provider + Send + Sync + 'static>(provider: T) -> Self {
+    fn new<T: Provider + Send + Sync + 'static>(provider: T, name: String) -> Self {
         Self {
             provider: Box::new(provider),
+            name,
         }
     }
 
@@ -135,6 +137,10 @@ impl ProviderTester {
             )
             .await?;
 
+        println!("=== {}::reponse1 ===", self.name);
+        dbg!(&response1);
+        println!("===================");
+
         // Verify we got a tool request
         assert!(
             response1
@@ -144,8 +150,6 @@ impl ProviderTester {
             "Expected tool request in response"
         );
 
-        dbg!(&response1);
-
         let id = &response1
             .content
             .iter()
@@ -178,7 +182,9 @@ impl ProviderTester {
             )
             .await?;
 
+        println!("=== {}::reponse2 ===", self.name);
         dbg!(&response2);
+        println!("===================");
 
         assert!(
             response2
@@ -216,6 +222,9 @@ where
     F: FnOnce() -> Result<T>,
     T: Provider + Send + Sync + 'static,
 {
+    // We start off as failed, so that if the process panics it is seen as a failure
+    TEST_REPORT.record_fail(name);
+
     // Take exclusive access to environment modifications
     let lock = ENV_LOCK.lock().unwrap();
 
@@ -276,7 +285,7 @@ where
         return Err(provider.err().expect("is error"));
     }
 
-    let tester = ProviderTester::new(provider.expect("already checked"));
+    let tester = ProviderTester::new(provider.expect("already checked"), name.to_string());
     match tester.run_test_suite().await {
         Ok(_) => {
             TEST_REPORT.record_pass(name);
@@ -328,7 +337,13 @@ async fn test_databricks_provider_oauth() -> Result<()> {
 
 #[tokio::test]
 async fn test_ollama_provider() -> Result<()> {
-    test_provider("Ollama", &[], None, ollama::OllamaProvider::from_env).await
+    test_provider(
+        "Ollama",
+        &["OLLAMA_HOST"],
+        None,
+        ollama::OllamaProvider::from_env,
+    )
+    .await
 }
 
 #[tokio::test]