Update app.py

app.py

@@ -416,6 +416,9 @@ body, .gradio-container {
     padding: 16px !important; color: #C0C3D0 !important;
     cursor: default !important; min-height: 120px !important;
 }
+#error-out {
+    width: 100% !important;
+}
 #error-out textarea {
     background: #1A0E0E !important;
     border: 1px solid #5a2020 !important;
@@ -427,6 +430,7 @@ body, .gradio-container {
     cursor: default !important;
     min-height: 120px !important;
     width: 100% !important;
+    box-sizing: border-box !important;
 }
 .divider { height: 1px; background: #2E3140; margin: 20px 0; }
 footer { display: none !important; }
@@ -475,16 +479,14 @@ def build_ui(evaluate_fn):
         accuracy, summary = _split_metrics(metrics)
     
         if error:
-            # Error occurred — show only the error box, hide all result boxes
             return (
-                gr.update(value="",       visible=False),  # verdict
-                gr.update(value="",       visible=False),  # accuracy
-                gr.update(value="",       visible=False),  # summary
-                gr.update(value="",       visible=False),  # issues
-                gr.update(value=error,    visible=True),   # error ← only this shows
+                gr.update(value="",  visible=True),           # verdict   — empty but visible
+                gr.update(value="",  visible=True),           # accuracy  — empty but visible
+                gr.update(value="",  visible=True),           # summary   — empty but visible
+                gr.update(value="",  visible=False),          # issues    — hidden
+                gr.update(value=error, visible=True),         # error     — shown
             )
     
-        # No error — show all result boxes, hide the error box
         if "PASS" in verdict.upper():
             verdict_display = "✅  PASS"
         elif "FAIL" in verdict.upper():
@@ -493,11 +495,11 @@ def build_ui(evaluate_fn):
             verdict_display = verdict or ""
     
         return (
-            gr.update(value=verdict_display, visible=True),   # verdict
-            gr.update(value=accuracy,        visible=True),   # accuracy
-            gr.update(value=summary,         visible=True),   # summary
-            gr.update(value=issues,          visible=True),   # issues
-            gr.update(value="",              visible=False),  # error ← hidden when no error
+            gr.update(value=verdict_display, visible=True),
+            gr.update(value=accuracy,        visible=True),
+            gr.update(value=summary,         visible=True),
+            gr.update(value=issues,          visible=True),   # issues    — shown
+            gr.update(value="",              visible=False),  # error     — hidden
         )
 
     with gr.Blocks(title="Python Code Evaluator") as demo:
@@ -555,7 +557,7 @@ def build_ui(evaluate_fn):
             error_out = gr.Textbox(
                 label="Status / Errors",
                 interactive=False,
-                visible=True,
+                visible=False,          # ← hidden by default, only shown on error
                 elem_id="error-out",
                 scale=1,
             )