[Discussion] Why not MappingProxyType?

[lifelonglearner127]

Thank you for writing this wonderful python package.
I don't think this is the right place to ask this question. However curious about this module. What are the differences from this package from types.MappingProxyType?
We can make the dictionary as read-only by using MappingProxyType?
Best Regards,
Levan Begashvili

[lurch]

I've never looked at MappingProxyType before, but https://stackoverflow.com/questions/41795116/difference-between-mappingproxytype-and-pep-416-frozendict might have some relevant info?

[lifelonglearner127]

Thanks for your quick reply.
so frozendict focus only on dict wherease MappingProxyType works as an interface, I think.
By the way, frozendict add immutability on the first-level just like MappingProxyType. So is it a known issue?

[Marco-Sulla]

I do not use MappingProxyType because is really slow. The old implementation used it. Since nothing stops you to change the old _dict attribute the contained the MappingProxyType, I didn't see any benefit from using it.

Py frozendict is not really immutable, since, well, it's written in Python. With this new version is only a bit more complicated, since you can change it only using object.__setattr__(). MappingProxyType is really immutable, but slow.

If you want a real immutable frozendict, you want the C Extension. If you have Linux you can install one of the Linux packages: https://github.com/Marco-Sulla/python-frozendict/releases If you don't have Linux, you can clone the repo and launch python3 setup.py c bdist_wheel and install the resulting wheel.

I would create a VM with CentOS 5 and create a manylinux distro, so I can publish it to Pypi. Furthermore, I have to remove the c parameter from setup.py and make it the default (if someone wants the pure Python only, a py parameters would be used instead)