Occurred 'cp949' error in IniConfig class

[melemo2]

@RonnyPfannschmidt

I solved that issue #37

f = open(self.path) -> f = open(self.path, encoding='utf-8')

[RonnyPfannschmidt]

@melemo2 please elaborate on the python version involved

im happy to add a sane or configriuable default, but im slightly hesitant to just put that in as is (we have had issues with people using their native language encoding instead of utf8 before)

[melemo2]

@RonnyPfannschmidt python version is 3.9.7 Detailed error msg and call stack is here https://github.com/scikit-learn/scikit-learn/issues/21933 I don't know why error occurred.. should I close this issue?

[RonnyPfannschmidt]

i believe this needs addressing, based on the links i'm also under the impression this happened to youu by accident as your editor use utf8 and your system encoding isnt

i should ensure we open files more safely later on

[melemo2]

@RonnyPfannschmidt thanks for your comment I think you're right :) It would be better to support both none and utf-8. by handling an 'UnicodeDecodeError' Exception

[Fengjun-Wang]

Hi @RonnyPfannschmidt and @melemo2 ,

If we can provide options on the encoding format, it could resolve problems for developer who are using Windows as development os. For example, as melemo2 mentioned, the scikit-learn has a "setup.cfg" file which is encoded as "utf8", however, some Windows os has default encoding cp936.

My idea is we can provide an additional param whose default value is utf-8, at https://github.com/pytest-dev/iniconfig/blob/master/src/iniconfig/__init__.py#L48 What do you think?

@melemo2 are you still working on MRG #37 ?

[melemo2]

@Fengjun-Wang

nope. I'll close it I mentioned problem already had been solved by just reinstall a project, so I'm not interested this issue.

but I think it's easier to handle exceptions than using parameter cuz If we add a parameter then another huge projects need to be modified :)