Core language¶
Classes¶
Special method __del__ not implemented for user-defined classes¶
Sample code:
import gc
class Foo:
def __del__(self):
print("__del__")
f = Foo()
del f
gc.collect()
| CPy output: | uPy output: |
__del__
|
Method Resolution Order (MRO) is not compliant with CPython¶
Cause: Depth first non-exhaustive method resolution order
Workaround: Avoid complex class hierarchies with multiple inheritance and complex method overrides. Keep in mind that many languages don’t support multiple inheritance at all.
Sample code:
class Foo:
def __str__(self):
return "Foo"
class C(tuple, Foo):
pass
t = C((1, 2, 3))
print(t)
| CPy output: | uPy output: |
Foo
|
(1, 2, 3)
|
When inheriting from multiple classes super() only calls one class¶
Cause: See Method Resolution Order (MRO) is not compliant with CPython
Workaround: See Method Resolution Order (MRO) is not compliant with CPython
Sample code:
class A:
def __init__(self):
print("A.__init__")
class B(A):
def __init__(self):
print("B.__init__")
super().__init__()
class C(A):
def __init__(self):
print("C.__init__")
super().__init__()
class D(B, C):
def __init__(self):
print("D.__init__")
super().__init__()
D()
| CPy output: | uPy output: |
D.__init__
B.__init__
C.__init__
A.__init__
|
D.__init__
B.__init__
A.__init__
|
Calling super() getter property in subclass will return a property object, not the value¶
Sample code:
class A:
@property
def p(self):
return {"a": 10}
class AA(A):
@property
def p(self):
return super().p
a = AA()
print(a.p)
| CPy output: | uPy output: |
{'a': 10}
|
<property>
|
Functions¶
Error messages for methods may display unexpected argument counts¶
Cause: Pycopy counts “self” as an argument.
Workaround: Interpret error messages with the information above in mind.
Sample code:
try:
[].append()
except Exception as e:
print(e)
| CPy output: | uPy output: |
append() takes exactly one argument (0 given)
|
function takes 2 positional arguments but 1 were given
|
Function objects do not have the __module__ attribute¶
Cause: MicroPython is optimized for reduced code size and RAM usage.
Workaround: Use sys.modules[function.__globals__['__name__']] for non-builtin modules.
Sample code:
def f():
pass
print(f.__module__)
| CPy output: | uPy output: |
__main__
|
Traceback (most recent call last):
File "<stdin>", line 13, in <module>
AttributeError: 'function' object has no attribute '__module__'
|
User-defined attributes for functions are not supported¶
Cause: Pycopy is highly optimized for memory usage.
Workaround: Use external dictionary, e.g. FUNC_X[f] = 0.
Sample code:
def f():
pass
f.x = 0
print(f.x)
| CPy output: | uPy output: |
0
|
Traceback (most recent call last):
File "<stdin>", line 13, in <module>
AttributeError: 'function' object has no attribute 'x'
|
Generator¶
Context manager __exit__() not called in a generator which does not run to completion¶
Sample code:
class foo(object):
def __enter__(self):
print("Enter")
def __exit__(self, *args):
print("Exit")
def bar(x):
with foo():
while True:
x += 1
yield x
def func():
g = bar(0)
for _ in range(3):
print(next(g))
func()
| CPy output: | uPy output: |
Enter
1
2
3
Exit
|
Enter
1
2
3
|
Runtime¶
Local variables aren’t included in locals() result¶
Cause: Pycopy doesn’t maintain symbolic local environment, it is optimized to an array of slots. Thus, local variables can’t be accessed by a name.
Sample code:
def test():
val = 2
print(locals())
test()
| CPy output: | uPy output: |
{'val': 2}
|
{'test': <function test at 0x7fffffff>, '__name__': '__main__', '__file__': '<stdin>'}
|
Code running in eval() function doesn’t have access to local variables¶
Cause: Pycopy doesn’t maintain symbolic local environment, it is optimized to an array of slots. Thus, local variables can’t be accessed by a name. Effectively, eval(expr) in Pycopy is equivalent to eval(expr, globals(), globals()).
Sample code:
val = 1
def test():
val = 2
print(val)
eval("print(val)")
test()
| CPy output: | uPy output: |
2
2
|
2
1
|
import¶
__path__ attribute of a package has a different type (single string instead of list of strings) in Pycopy¶
Cause: Pycopy does’t support namespace packages split across filesystem. Beyond that, Pycopy’s import system is highly optimized for minimal memory usage.
Workaround: Details of import handling is inherently implementation dependent. Don’t rely on such details in portable applications.
Sample code:
import modules
print(modules.__path__)
| CPy output: | uPy output: |
['/mnt/hdd/projects/pycopy/tests/cpydiff/modules']
|
../tests/cpydiff//modules
|
Failed to load modules are still registered as loaded¶
Cause: To make module handling more efficient, it’s not wrapped with exception handling.
Workaround: Test modules before production use; during development, use del sys.modules["name"], or just soft or hard reset the board.
Sample code:
import sys
try:
from modules import foo
except NameError as e:
print(e)
try:
from modules import foo
print("Should not get here")
except NameError as e:
print(e)
| CPy output: | uPy output: |
foo
name 'xxx' is not defined
foo
name 'xxx' is not defined
|
foo
name 'xxx' isn't defined
Should not get here
|
Pycopy does’t support namespace packages split across filesystem.¶
Cause: Pycopy’s import system is highly optimized for simplicity, minimal memory usage, and minimal filesystem search overhead.
Workaround: Don’t install modules belonging to the same namespace package in different directories. For Pycopy, it’s recommended to have at most 3-component module search paths: for your current application, per-user (writable), system-wide (non-writable).
Sample code:
import sys
sys.path.append(sys.path[1] + "/modules")
sys.path.append(sys.path[1] + "/modules2")
import subpkg.foo
import subpkg.bar
print("Two modules of a split namespace package imported")
| CPy output: | uPy output: |
Two modules of a split namespace package imported
|
Traceback (most recent call last):
File "<stdin>", line 13, in <module>
ImportError: no module named 'subpkg.bar'
|