warning message & checking for jit compilation

Author: Badr-MOUFAD

skglm/solvers/base.py

@@ -107,23 +107,28 @@ def solve(
         >>> ...
         >>> coefs, obj_out, stop_crit = solver.solve(X, y, datafit, penalty)
         """
-        # TODO do it properly instead of searching for a string
+        # TODO check for datafit/penalty being jit-compiled properly
+        # instead of searching for a string
         if "jitclass" in str(type(datafit)):
             warnings.warn(
-                "Do not pass a compiled datafit, compilation is done inside solver now"
+                "Passing in a compiled datafit is deprecated since skglm v0.5 "
+                "Compilation is now done inside solver."
+                "This will raise an error starting skglm v0.6 onwards."
             )
+        elif datafit is not None:
+            datafit = compiled_clone(datafit, to_float32=X.dtype == np.float32)
+
         if "jitclass" in str(type(penalty)):
             warnings.warn(
-                "Do not pass a compiled penalty, compilation is done inside solver now"
+                "Passing in a compiled penalty is deprecated since skglm v0.5 "
+                "Compilation is now done inside solver. "
+                "This will raise an error starting skglm v0.6 onwards."
             )
-        else:
-            if datafit is not None:
-                datafit = compiled_clone(datafit, to_float32=X.dtype == np.float32)
-            if penalty is not None:
-                penalty = compiled_clone(penalty)
-                # TODO add support for bool spec in compiled_clone
-                # currently, doing so break the code
-                # penalty = compiled_clone(penalty, to_float32=X.dtype == np.float32)
+        elif penalty is not None:
+            penalty = compiled_clone(penalty)
+            # TODO add support for bool spec in compiled_clone
+            # currently, doing so break the code
+            # penalty = compiled_clone(penalty, to_float32=X.dtype == np.float32)
 
         if run_checks:
             self._validate(X, y, datafit, penalty)