int.__sizeof__ from #1426
#1541
Conversation
vm/src/obj/objint.rs
Outdated
| @@ -186,6 +186,11 @@ impl PyInt { | |||
| PyInt::new(options.get_int_value(vm)?).into_ref_with_type(vm, cls) | |||
| } | |||
|
|
|||
| #[pymethod(name = "__sizeof__")] | |||
| fn sizeof(&self, _vm: &VirtualMachine) -> usize { | |||
| (self.value.bits() as f64 / 8.).ceil() as usize | |||
There was a problem hiding this comment.
Results in:
>>>>> (2**1).__sizeof__()
1
>>>>> (2**8).__sizeof__()
2
>>>>> (2**16).__sizeof__()
3
>>>>> (2**32).__sizeof__()
5
>>>>> (2**64).__sizeof__()
9Edit: tested better values @coolreader18
vm/src/obj/objint.rs
Outdated
| @@ -188,7 +189,7 @@ impl PyInt { | |||
|
|
|||
| #[pymethod(name = "__sizeof__")] | |||
| fn sizeof(&self, _vm: &VirtualMachine) -> usize { | |||
| (self.value.bits() as f64 / 8.).ceil() as usize | |||
| mem::size_of::<Self>() as usize | |||
There was a problem hiding this comment.
Results in:
>>>>> for i in range(0, 100, 10):
..... print((i).__sizeof__())
.....
32
32
32
.
.
.
32
32
mem::size_of::<Self>() as usizeThis'll just always return 32 I believe, I'm not sure how useful this is. You're right that it goes deeper than I thought 😅 |
|
Yeah, it's not as simple as |
|
Although I don't think there was, it just didn't go above 1 byte because 100 can be contained with just 8 binary digits; you'd have to go above 256 to see a change. |
vm/src/obj/objint.rs
Outdated
| @@ -189,7 +188,7 @@ impl PyInt { | |||
|
|
|||
| #[pymethod(name = "__sizeof__")] | |||
| fn sizeof(&self, _vm: &VirtualMachine) -> usize { | |||
| mem::size_of::<Self>() as usize | |||
| self.value.bits() + 7 & !7 | |||
There was a problem hiding this comment.
>>>>> (2**1).__sizeof__()
8
>>>>> (2**8).__sizeof__()
16
>>>>> (2**16).__sizeof__()
24
>>>>> (2**32).__sizeof__()
40There was a problem hiding this comment.
@coolreader18 I think you're right about just not going high enough, I'll test that out
There was a problem hiding this comment.
sizeof int is sum of size of int object itself + sum of its heap allocated memory.
So, roughly, mem::sizeof<T> + 4 * self.value.data.len() is expected.
If current code describes the latter, adding mem::sizeof will make the value correct
vm/src/obj/objint.rs
Outdated
| @@ -188,7 +189,7 @@ impl PyInt { | |||
|
|
|||
| #[pymethod(name = "__sizeof__")] | |||
| fn sizeof(&self, _vm: &VirtualMachine) -> usize { | |||
| self.value.bits() + 7 & !7 | |||
| mem::size_of::<Self>() + (self.value.bits() + 7) & !7 | |||
There was a problem hiding this comment.
I think this should be 32 bits on the stack + bits on the heap rounded up to the nearest 8 bits.
>>>>> (2**2).__sizeof__()
40
>>>>> (2**8).__sizeof__()
48
>>>>> (2**16).__sizeof__()
56
>>>>> (2**32).__sizeof__()
72There was a problem hiding this comment.
__sizeof__ should actually return bytes, so you'd probably want to divide the bits you get by 8
5015024 to
8fff494
Compare
vm/src/obj/objint.rs
Outdated
| @@ -576,7 +576,7 @@ impl PyInt { | |||
|
|
|||
| #[pymethod(name = "__sizeof__")] | |||
| fn sizeof(&self, _vm: &VirtualMachine) -> usize { | |||
| size_of::<Self>() | |||
| (size_of::<Self>() + (self.value.bits() + 7) & !7) / 8 | |||
There was a problem hiding this comment.
Divided by 8 to return bytes instead of bits.
There was a problem hiding this comment.
mem::size_of actually does return bytes, it's the BigInt::bits method that returns bits.
Co-Authored-By: Noah <33094578+coolreader18@users.noreply.github.com>
Impl from @RobertBerglund 's comment, working on #304