[Lang] Raise errors when using the packed switch (#7125)

### Brief Summary

The `packed` switch of `ti.init()` should be removed according to the
[deprecation
notice](https://github.com/taichi-dev/taichi/releases/tag/v1.3.0).
However, to avoid a large changeset immediately before release (#7104),
let's raise errors first and get that in later.

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

docs/lang/articles/contribution/write_test.md

@@ -283,5 +283,4 @@ Now, Taichi supports the following extensions:
 | bls           | Block-local storage                                           |
 | assertion     | Run-time asserts in Taichi kernels                            |
 | extfunc       | Support inserting external function calls or backend source   |
-| packed        | Packed mode: shapes will not be padded to powers of two       |
 | dynamic_index | Dynamic index support for tensors                             |

python/taichi/lang/misc.py

@@ -9,6 +9,7 @@
 from taichi._lib import core as _ti_core
 from taichi._lib.utils import locale_encode
 from taichi.lang import impl
+from taichi.lang.exception import TaichiRuntimeError
 from taichi.lang.expr import Expr
 from taichi.lang.impl import axes, get_runtime
 from taichi.profiler.kernel_profiler import get_default_kernel_profiler
@@ -184,7 +185,7 @@
 
 The list of currently available extensions is ['sparse', 'quant', \
     'mesh', 'quant_basic', 'data64', 'adstack', 'bls', 'assertion', \
-        'extfunc', 'packed', 'dynamic_index'].
+        'extfunc', 'dynamic_index'].
 """
 
 
@@ -346,7 +347,6 @@ def init(arch=None,
             * ``cpu_max_num_threads`` (int): Sets the number of threads used by the CPU thread pool.
             * ``debug`` (bool): Enables the debug mode, under which Taichi does a few more things like boundary checks.
             * ``print_ir`` (bool): Prints the CHI IR of the Taichi kernels.
-            * ``packed`` (bool): Enables the packed memory layout. See https://docs.taichi-lang.org/docs/layout.
             *``offline_cache`` (bool): Enables offline cache of the compiled kernels. Default to True. When this is enabled Taichi will cache compiled kernel on your local disk to accelerate future calls.
             *``random_seed`` (int): Sets the seed of the random generator. The default is 0.
     """
@@ -362,15 +362,7 @@ def init(arch=None,
         check_require_version(require_version)
 
     if "packed" in kwargs:
-        if kwargs["packed"] is True:
-            warnings.warn(
-                "Currently packed=True is the default setting and the switch will be removed in v1.4.0.",
-                DeprecationWarning)
-        else:
-            warnings.warn(
-                "The automatic padding mode (packed=False) will no longer exist in v1.4.0. The switch will "
-                "also be removed then. Make sure your code doesn't rely on it.",
-                DeprecationWarning)
+        raise TaichiRuntimeError("The 'packed' switch has been removed.")
 
     if "dynamic_index" in kwargs:
         warnings.warn(

tests/python/test_bitmasked.py

@@ -2,7 +2,8 @@
 from tests import test_utils
 
 
-def _test_basic():
+@test_utils.test(require=ti.extension.sparse)
+def test_basic():
     x = ti.field(ti.i32)
     c = ti.field(ti.i32)
     s = ti.field(ti.i32)
@@ -30,16 +31,6 @@ def sum():
     assert s[None] == 42
 
 
-@test_utils.test(require=ti.extension.sparse)
-def test_basic():
-    _test_basic()
-
-
-@test_utils.test(require=ti.extension.sparse, packed=True)
-def test_basic_packed():
-    _test_basic()
-
-
 @test_utils.test(require=ti.extension.sparse)
 def test_bitmasked_then_dense():
     x = ti.field(ti.f32)
@@ -177,7 +168,8 @@ def deactivate():
     assert c[None] == 0
 
 
-def _test_sparsity_changes():
+@test_utils.test(require=ti.extension.sparse)
+def test_sparsity_changes():
     x = ti.field(ti.i32)
     c = ti.field(ti.i32)
     s = ti.field(ti.i32)
@@ -210,16 +202,6 @@ def run():
     assert s[None] == 42
 
 
-@test_utils.test(require=ti.extension.sparse)
-def test_sparsity_changes():
-    _test_sparsity_changes()
-
-
-@test_utils.test(require=ti.extension.sparse, packed=True)
-def test_sparsity_changes_packed():
-    _test_sparsity_changes()
-
-
 @test_utils.test(require=ti.extension.sparse)
 def test_bitmasked_offset_child():
     x = ti.field(ti.i32)

tests/python/test_deprecation.py

@@ -93,22 +93,12 @@ def test_deprecate_metal_sparse():
         ti.root.dynamic(ti.i, 10)
 
 
-def test_deprecated_packed_true():
-    with pytest.warns(
-            DeprecationWarning,
-            match=
-            "Currently packed=True is the default setting and the switch will be removed in v1.4.0."
-    ):
-        ti.init(packed=True)
-
-
-def test_deprecated_packed_false():
-    with pytest.warns(
-            DeprecationWarning,
-            match=
-            r"The automatic padding mode \(packed=False\) will no longer exist in v1.4.0. The switch will "
-            "also be removed then. Make sure your code doesn't rely on it."):
-        ti.init(packed=False)
+# Remove this before v1.5.0
+@pytest.mark.parametrize("value", [True, False])
+def test_removed_packed(value):
+    with pytest.raises(ti.TaichiRuntimeError,
+                       match="The 'packed' switch has been removed."):
+        ti.init(packed=value)
 
 
 @test_utils.test(arch=ti.vulkan)

tests/python/test_mpm_particle_list.py

@@ -54,14 +54,3 @@ def test_mpm_particle_list_no_leakage():
     # By default Taichi will allocate 0.5 GB for testing.
     mpm = MPMSolver(res=(128, 128))
     mpm.step()
-
-
-@pytest.mark.run_in_serial
-@test_utils.test(require=ti.extension.sparse,
-                 exclude=[ti.metal],
-                 device_memory_GB=1.0,
-                 packed=True)
-def test_mpm_particle_list_no_leakage_packed():
-    # By default Taichi will allocate 0.5 GB for testing.
-    mpm = MPMSolver(res=(128, 128))
-    mpm.step()

tests/python/test_packed_size.py

@@ -2,7 +2,7 @@
 from tests import test_utils
 
 
-@test_utils.test(arch=[ti.cpu, ti.cuda], packed=True)
+@test_utils.test(arch=[ti.cpu, ti.cuda])
 def test_packed_size():
     x = ti.field(ti.i32)
     ti.root.dense(ti.l, 3).dense(ti.ijk, 129).place(x)

tests/python/test_sparse_basics.py

@@ -74,7 +74,8 @@ def func():
     assert s[None] == 3
 
 
-def _test_pointer2():
+@test_utils.test(require=ti.extension.sparse)
+def test_pointer2():
     x = ti.field(ti.f32)
     s = ti.field(ti.i32)
 
@@ -103,16 +104,6 @@ def func():
     assert s[None] == 5 * n
 
 
-@test_utils.test(require=ti.extension.sparse)
-def test_pointer2():
-    _test_pointer2()
-
-
-@test_utils.test(require=ti.extension.sparse, packed=True)
-def test_pointer2_packed():
-    _test_pointer2()
-
-
 @pytest.mark.skip(reason='https://github.com/taichi-dev/taichi/issues/2520')
 @test_utils.test(require=ti.extension.sparse)
 def test_pointer_direct_place():

tests/python/test_struct_for_intermediate.py

@@ -21,25 +21,11 @@ def iterate():
             assert x[i * n, j * m] == 1, (i, j)
 
 
-@test_utils.test(require=ti.extension.sparse,
-                 demote_dense_struct_fors=False,
-                 packed=False)
+@test_utils.test(require=ti.extension.sparse, demote_dense_struct_fors=False)
 def test_nested():
     _test_nested()
 
 
-@test_utils.test(demote_dense_struct_fors=True, packed=False)
+@test_utils.test(demote_dense_struct_fors=True)
 def test_nested_demote():
     _test_nested()
-
-
-@test_utils.test(require=ti.extension.sparse,
-                 demote_dense_struct_fors=False,
-                 packed=True)
-def test_nested_packed():
-    _test_nested()
-
-
-@test_utils.test(demote_dense_struct_fors=True, packed=True)
-def test_nested_demote_packed():
-    _test_nested()

tests/python/test_struct_for_non_pot.py

@@ -2,7 +2,8 @@
 from tests import test_utils
 
 
-def _test_1d():
+@test_utils.test()
+def test_1d():
     x = ti.field(ti.i32)
     sum = ti.field(ti.i32)
 
@@ -23,16 +24,7 @@ def accumulate():
 
 
 @test_utils.test()
-def test_1d():
-    _test_1d()
-
-
-@test_utils.test(packed=True)
-def test_1d_packed():
-    _test_1d()
-
-
-def _test_2d():
+def test_2d():
     x = ti.field(ti.i32)
     sum = ti.field(ti.i32)
 
@@ -58,17 +50,8 @@ def accumulate():
         assert sum[None] == gt
 
 
-@test_utils.test()
-def test_2d():
-    _test_2d()
-
-
-@test_utils.test(packed=True)
-def test_2d_packed():
-    _test_2d()
-
-
-def _test_2d_pointer():
+@test_utils.test(require=ti.extension.sparse)
+def test_2d_pointer():
     block_size, leaf_size = 3, 8
     x = ti.field(ti.i32)
     block = ti.root.pointer(ti.ij, (block_size, block_size))
@@ -93,13 +76,3 @@ def test() -> ti.i32:
             ans += i + j * 2
 
     assert ans == test()
-
-
-@test_utils.test(require=ti.extension.sparse, packed=False)
-def test_2d_pointer():
-    _test_2d_pointer()
-
-
-@test_utils.test(require=ti.extension.sparse, packed=True)
-def test_2d_pointer_packed():
-    _test_2d_pointer()