Bump pylint (#408)

This patch bumps pylint to the latest version. This is in preparation
for bumping the rest of the deps.

.pylintrc

@@ -104,12 +104,14 @@ disable=abstract-method,
         no-self-use,
 	no-value-for-parameter, # gin causes false positives
         nonzero-method,
+        not-callable, # lots of false positives
         oct-method,
         old-division,
         old-ne-operator,
         old-octal-literal,
         old-raise-syntax,
         parameter-unpacking,
+        possibly-used-before-assignment, # false positives with control flow
         print-statement,
         raising-string,
         range-builtin-not-iterating,
@@ -132,6 +134,7 @@ disable=abstract-method,
         too-many-instance-attributes,
         too-many-locals,
         too-many-nested-blocks,
+        too-many-positional-arguments,
         too-many-public-methods,
         too-many-return-statements,
         too-many-statements,
@@ -140,6 +143,7 @@ disable=abstract-method,
         unicode-builtin,
         unnecessary-pass,
         unpacking-in-except,
+        used-before-assignment,
         useless-else-on-loop,
         useless-object-inheritance,
         useless-suppression,
@@ -425,7 +429,7 @@ valid-metaclass-classmethod-first-arg=mcs
 
 # Exceptions that will emit a warning when being caught. Defaults to
 # "Exception"
-overgeneral-exceptions=StandardError,
-                       Exception,
-                       BaseException
+overgeneral-exceptions=builtins.StandardError,
+                       builtins.Exception,
+                       builtins.BaseException
 

Pipfile

@@ -12,7 +12,7 @@ tensorflow = "==2.12.0"
 dm-reverb = "==0.11.0"
 
 [dev-packages]
-pylint = "==2.14.1"
+pylint = "==3.3.2"
 pytest = "==7.1.2"
 pytype = "==2022.06.06"
 yapf = "==0.43.0"

compiler_opt/distributed/buffered_scheduler_test.py

@@ -63,16 +63,18 @@ def square(self, the_value, extra_factor=1):
                            [x * x for x in range(10)])
 
       _, futures = buffered_scheduler.schedule_on_worker_pool(
-          lambda w, v: w.square(**v), [dict(the_value=v) for v in range(10)],
-          pool)
+          lambda w, v: w.square(**v), [{
+              'the_value': v
+          } for v in range(10)], pool)
       worker.wait_for(futures)
       self.assertListEqual([f.result() for f in futures],
                            [x * x for x in range(10)])
 
       # same idea, but mix some kwargs
       _, futures = buffered_scheduler.schedule_on_worker_pool(
-          lambda w, v: w.square(v[0], **v[1]),
-          [(v, dict(extra_factor=10)) for v in range(10)], pool)
+          lambda w, v: w.square(v[0], **v[1]), [(v, {
+              'extra_factor': 10
+          }) for v in range(10)], pool)
       worker.wait_for(futures)
       self.assertListEqual([f.result() for f in futures],
                            [x * x * 10 for x in range(10)])

compiler_opt/distributed/worker_test.py

@@ -35,8 +35,11 @@ def test_gin_args(self):
         gin.bind_parameter('_test.SomeType.argument', 42)
       real_args = worker.get_full_worker_args(
           SomeType, more_args=2, even_more_args='hi')
-      self.assertDictEqual(real_args,
-                           dict(argument=42, more_args=2, even_more_args='hi'))
+      self.assertDictEqual(real_args, {
+          'argument': 42,
+          'more_args': 2,
+          'even_more_args': 'hi'
+      })
 
 
 if __name__ == '__main__':

compiler_opt/es/blackbox_learner.py

@@ -168,7 +168,7 @@ def _get_perturbations(self) -> List[npt.NDArray[np.float32]]:
     rng = np.random.default_rng(seed=self._seed)
     for _ in range(self._config.total_num_perturbations):
       perturbations.append(
-          rng.normal(size=(len(self._model_weights))) *
+          rng.normal(size=len(self._model_weights)) *
           self._config.precision_parameter)
     return perturbations
 

compiler_opt/es/blackbox_optimizers.py

@@ -983,16 +983,14 @@ def trust_region_test(self, current_input: FloatArray,
                      str(tr_imp_ratio))
       if should_reject:
         self.radius *= self.params['reject_factor']
-        if self.radius < self.params['minimum_radius']:
-          self.radius = self.params['minimum_radius']
+        self.radius = max(self.radius, self.params['minimum_radius'])
         self.is_returned_step = True
         print('Step rejected. Shrink: ' + str(self.radius) + log_message)
         return False
       else:  # accept step
         if should_shrink:
           self.radius *= self.params['shrink_factor']
-          if self.radius < self.params['minimum_radius']:
-            self.radius = self.params['minimum_radius']
+          self.radius = max(self.radius, self.params['minimum_radius'])
           print('Shrink: ' + str(self.radius) + log_message)
         elif should_grow:
           self.radius *= self.params['grow_factor']

compiler_opt/rl/compilation_runner.py

@@ -122,15 +122,15 @@ def __init__(self):
     Exception.__init__(self)
 
 
-def kill_process_ignore_exceptions(p: 'subprocess.Popen[bytes]'):
+def kill_process_ignore_exceptions(p: 'subprocess.Popen[bytes]'):  # pylint: disable=useless-return
   # kill the process and ignore exceptions. Exceptions would be thrown if the
   # process has already been killed/finished (which is inherently in a race
   # condition with us killing it)
   try:
     p.kill()
     p.wait()
   finally:
-    return  # pylint: disable=lost-exception
+    return  # pylint: disable=lost-exception,return-in-finally
 
 
 class WorkerCancellationManager:

compiler_opt/rl/local_data_collector.py

@@ -105,13 +105,12 @@ def _schedule_jobs(self, policy: policy_saver.Policy, model_id: int,
     # by now, all the pending work, which was signaled to cancel, must've
     # finished
     self._join_pending_jobs()
-    jobs = [
-        dict(
-            loaded_module_spec=loaded_module_spec,
-            policy=policy,
-            reward_stat=self._reward_stat_map[loaded_module_spec.name],
-            model_id=model_id) for loaded_module_spec in sampled_modules
-    ]
+    jobs = [{
+        'loaded_module_spec': loaded_module_spec,
+        'policy': policy,
+        'reward_stat': self._reward_stat_map[loaded_module_spec.name],
+        'model_id': model_id
+    } for loaded_module_spec in sampled_modules]
 
     (self._workers,
      self._current_futures) = buffered_scheduler.schedule_on_worker_pool(

compiler_opt/type_map.py

@@ -17,11 +17,10 @@
 from typing import List, Tuple, Union
 import tensorflow as tf
 
-ScalarCType = Union['type[ctypes.c_float]', 'type[ctypes.c_double]',
-                    'type[ctypes.c_int8]', 'type[ctypes.c_int16]',
-                    'type[ctypes.c_uint16]', 'type[ctypes.c_int32]',
-                    'type[ctypes.c_uint32]', 'type[ctypes.c_int64]',
-                    'type[ctypes.c_uint64]']
+ScalarCType = Union[  # pylint: disable=invalid-name
+    'type[ctypes.c_float]', 'type[ctypes.c_double]', 'type[ctypes.c_int8]',
+    'type[ctypes.c_int16]', 'type[ctypes.c_uint16]', 'type[ctypes.c_int32]',
+    'type[ctypes.c_uint32]', 'type[ctypes.c_int64]', 'type[ctypes.c_uint64]']
 
 TYPE_ASSOCIATIONS: List[Tuple[str, ScalarCType,
                               tf.DType]] = [