Provide python bindings

[vpereira]

Hi, since the python-blkmaker is outdated (and it looks unmaintained), wouldn't be a good idea to add python bindings to the libblkmaker?

[luke-jr]

I've been pondering this for a week or so, actually, but I'm not sure if C code with Python bindings is sufficiently usable in Python code, or if python-blkmaker should be updated. Nor do I have any experience with writing Python bindings that get installed properly. (Is this something perhaps you could contribute?)

[vpereira]

I was looking into that right now... Normally You don't have to write python at all (http://intermediate-and-advanced-software-carpentry.readthedocs.io/en/latest/c++-wrapping.html) I wrote in the past some FFI classes and ctype interfaces to access C code from python code.. Let's see what I can do with Pyrex..

Or maybe I can help updating the python-blkmaker.. what do you prefer?

BTW, this python "miner" getblocktemplate implementation looks good and works even with testnet (all tested with bitcoind 0.12.1) https://github.com/vsergeev/ntgbtminer/blob/master/ntgbtminer.py

[luke-jr]

IMO in an ideal world, there would be a pure-Python python-blkmaker and also libblkmaker Python bindings that are interchangeable, but that may take too much time to maintain.

The main advantage of python-blkmaker is that users/developers don't need to compile a C library. The main advantage of bindings would be a more common codebase.

Gentoo downstream tells me the preferred way to do bindings would be a separate repo for the bindings themselves, using distutils/setuptools for their install.

Updating python-blkmaker at this point would require porting:

• a221293 Refactor construction of block submissions to allow for results from get_mdata
• e3465d3 Refactor blkmk_assemble_submission2_ to work with either extranonce/mdata or dataid/data
• 58aa6a3 Refactor to eliminate blkmk_assemble_submission_ wrapper
• 7bf7e80 Bugfix: Expose blkmk_submitm_jansson for submitting results for blkmk_get_mdata
• c94e438 Bugfix: Always avoid data/mdata size conflict in blkmk_append_coinbase_safe2 since nothing else checks merkle_only (nor has it available to check)

I've pushed a more recently updated python-blkmaker to https://github.com/luke-jr/python-blkmaker for now.

[luke-jr]

There, finished updating python-blkmaker...

[vpereira]

great! One advantage to have the Python wrapper is to use it to unit test the whole lib. Unit test in python is way easier than in C.

[luke-jr]

Yes, I'd still welcome Python bindings for that reason. :)