on rc tag project should be uploaded to https://test.pypi.org/ and tested.
on release tag project should be uploaded to https://pypi.org/

I suggest to do it using github actions

We could add (in CI) https://github.com/PyCQA/bandit scan to check for known vulnerabilities.

FEAT: Implement iterators API

Rationale

Easier to use and more natural API.

Description

Implementation based on C++ iterators.
https://pmem.io/pmemkv/master/doxygen/classpmem_1_1kv_1_1db_1_1iterator.html

API Changes

New Iterator class.

Implementation details

Meta

ref: https://fedoramagazine.org/make-your-python-code-look-good-with-black-on-fedora/

Numpy would be a good example for pmemkv where we can do some operations on huge values stored in persistent memory. Numpy evaluates (loads) data lazily which means that it would greatly benefit from direct access to pmem.

This example could also be used to verify whether we really allow for direct from python (check whether there are no copies made).

add automatic generation of gh-pages docs, using sphinx

and additionally, link to pmem.io

We can get rid of statuses and use exceptions instead. Perhaps the only method(s) to return something will be get method(s).

at least few tweaks/bugs to address:

• as @karczex suggested we could move from distutils to setuptools
• it seems that custom pmemkv.so paths (inlcude and lib) are not taken into account
• some data in package is missing (see below) - need to setup them, as it is done e.g. in other bindings

Home-page: UNKNOWN
Author: UNKNOWN
Author-email: UNKNOWN
License: UNKNOWN
Description: UNKNOWN
Platform: UNKNOWN

pmemkv seems to throw only general Exception ('PyExc_Exception') which is not very convenient in handling. Some more descriptive built-in exceptions may be thrown instead, see: https://docs.python.org/3.7/library/exceptions.html