Integers do not get casted in operations with complex variables

[faze-geek]

While playing with the complex module I saw certain expressions giving incorrect results .In Python, when an integer is operated with objects of complex() class the integer is implicitly casted to a complex variable with 0 imaginary value.

y = complex(5)+100
print(y)
(lp) C:\Users\kunni\lpython>python try.py
(105+0j)

(lp) C:\Users\kunni\lpython>src\bin\lpython try.py
(5.000000,0.000000)

y = complex(5)*10
print(y)
(lp) C:\Users\kunni\lpython>python try.py
(50+0j)

(lp) C:\Users\kunni\lpython>src\bin\lpython try.py
(0.000000,0.000000)

y = c32(5)-10
print(y)
(lp) C:\Users\kunni\lpython>python try.py
(-5+0j)

(lp) C:\Users\kunni\lpython>src\bin\lpython try.py
(5.000000,-0.000000)

[faze-geek]

All the operations work correctly with float variables -

def main1():
    y:c32
    y = c32(5)*1.5
    print(y)
    y = c32(5)+1.5
    print(y)
    y = complex(5)-1.5
    print(y)
    y = c32(5)/1.5
    print(y)


main1()

(lp) C:\Users\kunni\lpython>src\bin\lpython try.py
(7.500000,0.000000)
(6.500000,0.000000)
(3.500000,-0.000000)
(3.333333,0.000000)

[Smit-create]

Thanks, I'll have a look.

[Smit-create]

Are you trying this on the main branch?

It works fine for me:

from ltypes import c32

def f():
    y: c32
    y = complex(5)+c32(100)
    print(y)
    y = complex(5)*c32(10)
    print(y)

f()

LPython:

% lpython a.py
(105.000000,0.000000)
(50.000000,0.000000)

CPython:

% PYTHONPATH=$PWD/src/runtime/ltypes python a.py 
(105+0j)
(50+0j)

[faze-geek]

Are you trying this on the main branch?
It works fine for me:

Yes, I cannot confirm right now but I think the example you are writing worked with the main branch for me as well, it's when you do complex(5) + 100 when the result is incorrect.

[faze-geek]

Could you run the exact examples which I have given in the issue and recheck ? I had typed y: c32 above, rest everything is shown.

[Smit-create]

That throws some other error:

from ltypes import c32

def f():
    y: c32
    y = complex(5)+100
    print(y)

f()

Output:

case exprType::BinOp: { v.visit_BinOp((const BinOp_t &)x); return; }
  File "/Users/thebigbool/repos/lpython/src/lpython/semantics/python_ast_to_asr.cpp", line 2813
    }
  File "/Users/thebigbool/repos/lpython/src/lpython/semantics/python_ast_to_asr.cpp", line 2091
    ASRUtils::expr_value(right));
  File "/Users/thebigbool/repos/lpython/src/libasr/asr.h", line 40
    LCOMPILERS_ASSERT(is_a<T>(*f));
AssertFailed: is_a<T>(*f)

Can you provide the exact snippet you tried and also please make sure you are using main branch?

[faze-geek]

Can you provide the exact snippet you tried and also please make sure you are using main branch?

Sure I'll do that soon . Thankfully the snippet leads to an error only, so raising this issue wasn't a waste !

[faze-geek]

Yes those above results were run on the main branch. I checked now.
After pulling changes from #1525 I get the following -

def main1():
    y:c32
    y = c32(5)*1
    print(y)

main()

(lp) C:\Users\kunni\lpython>src\bin\lpython try.py
semantic error: Type mismatch in binary operator; the types must be compatible
 --> try.py:3:9
  |
3 |     y = c32(5)*1
  |         ^^^^^^ ^ type mismatch (c32 and i32)

And same incompatibility is shown by all binary operators. Can someone confirm this once ? I am using a windows machine
(maybe the cause of different output earlier ??)

[Smit-create]

That is the output we expect.

[faze-geek]

That is the output we expect.

Hey @Smit-create
I still have my doubts. Shouldn't the output be inline with that of Cpython ?

y = complex(5)+100
print(y)
(lp) C:\Users\kunni\lpython>python try.py
(105+0j)

(lp) C:\Users\kunni\lpython>src\bin\lpython try.py
(5.000000,0.000000)

y = complex(5)*10
print(y)
(lp) C:\Users\kunni\lpython>python try.py
(50+0j)

(lp) C:\Users\kunni\lpython>src\bin\lpython try.py
(0.000000,0.000000)

If it is still what is expected I'll close this issue soon. Thanks !

[Smit-create]

I still have my doubts. Shouldn't the output be inline with that of Cpython ?

Can you please provide the complete python snippet you are trying so that I can reproduce the same results? I am unable to get these results on the main branch.