propose fix for issue 61

[zh3nl]

Fix: For fget, fset, and fdel, I suggest binding the relevant methods to the class. Also use a descriptor protocol for class members.

Tests: I have created the following tests to test my implementation against read-only class property, deletable class property, class properties with default values, and class property with custom getter/setter methods.

Please provide feedback as I am a beginner to this!

[VigneshVSV]

I will also do a step-by-step debug and get back in about a day! Thanks a lot.

[VigneshVSV]

Hi, I think we have to rever the latest commit because some other sections of the code are breaking by changing getter/setter decorator to return self.

I have made a separate issue here: #63

[VigneshVSV]

Also the version in here and here can be changed to 0.2.10 so that once this is merged, I can release it to pip and anaconda.

[zh3nl]

@VigneshVSV Just pushed the following changes:

1. Reverted changes to decorators to the original commit
2. Asserted instance-based access as you suggested.
3. Added self.assertEqual(instance.deletable_class_prop, 100)

[VigneshVSV]

To run the full unit tests, you can do

cd tests
python -m unittest

It needs to pass before I can merge the branch. Unfortunately , for non-admin, its not automatically triggered.

[zh3nl]

I have attempted to change my code to pass the unit tests but I observe that the set and del methods are not working because they are trying to change the class which will result in "TypeError: 'mappingproxy' object does not support item assignment". This may or may not be in the scope of the issue and I do not want to change too much of the code.

[VigneshVSV]

I have attempted to change my code to pass the unit tests but I observe that the set and del methods are not working because they are trying to change the class which will result in "TypeError: 'mappingproxy' object does not support item assignment". This may or may not be in the scope of the issue and I do not want to change too much of the code.

Not a problem!

This snippet should fix that

else: 
    if self.class_member:
        setattr(obj, self._internal_name, value)
    else:
        obj.__dict__[self._internal_name] = value

However, I can take it from here. Should I do so? I think other parts are working when I checked also in the debugger.

If you interested in an explanation, I think this is due class-factory or metaclass behaviour where the __dict__ seems to be implemented as a mappingproxy. It needs a setattr operation instead of using it like a dictionary.

[VigneshVSV]

Also, thanks a lot for the contribution!

[zh3nl]

I have implemented the changes you suggested. If you want to take it from here, please do so and merge my additions as needed.

[VigneshVSV]

I have implemented the changes you suggested. If you want to take it from here, please do so and merge my additions as needed.

Thanks again! Its looking pretty good.

[zh3nl]

I want to note that in #62 in the chunk between 413 and 420 on parametrized.py I did use getattr rather than obj.dict.get. Not sure if will affect anything.

[VigneshVSV]

I want to note that in #62 in the chunk between 413 and 420 on parametrized.py I did use getattr rather than obj.dict.get. Not sure if will affect anything.

Logically its fine! Its probably just a mere few nanoseconds faster to just use the __dict__. The getattr will ultimately look into the __dict__ if it does not find it anywhere else.

[zh3nl]

Sounds good! I thought there was a slight difference between the two, but wanted to make sure that I wouldn't break anything :) Have a great rest of your day, and thanks for letting me contribute to the project!

[VigneshVSV]

Hi, I published it to pip. Thanks a lot!

I had to make one change which was, I reset self.fget.__get__(None, obj)(obj) to self.fget(obj) & similar for setters and deleters because when you try to bind it using None as the first argument, it results in the same object.
In other words self.fget.__get__(None, obj) is exactly equal to self.fget(obj).

You can have a look here: https://pypi.org/project/hololinked/

[zh3nl]

Okay got it. So this change has the same functionality but is more readable?

[VigneshVSV]

Okay got it. So this change has the same functionality but is more readable?

It just avoids one extra function call. It does not make it more or less readable as if you try self.fget.__get__(None, cls) == self.fget, you will get True.