the future of happybase

[wbolster]

considering that:

• happybase is quite a mature and stable library
• happybase is widely used
• happybase has not seen much development lately
• the original author (me, @wbolster) is not very active as a maintainer anymore
• various pull requests and bug reports have not received proper attention
• @aiudirog has recently done fantastic work on making an async version of happybase: aiohappybase
• the same aiohappybase work modernizes the code to be python3.6+ only
• the EOL dates of python2 and python3.5 are in the near future

it may be good for happybase's future if:

• others take over (part of) the project's maintenance
• happybase drops python 2 support and becomes python3.6+
• happybase can benefit from modern python (type annotations + mypy, async syntax)
• merges with the aiohappybase fork if it becomes a strict superset of the original happybase, which seems technically feasible (https://github.com/aiudirog/aiohappybase/pull/1, https://github.com/aiudirog/aiohappybase/pull/2)

thoughts welcome, especially from @aiudirog (with whom i also have had recent email contact).

[wbolster]

i've transferred my wbolster/happybase github repository into a new python-happybase github organisation (unfortunately happybase was taken already), and i've invited @aiudirog to become a co-owner of this new github org.

[aiudirog]

I should have my third idea for merging the two forks (where async is the default) fleshed out by tomorrow.

With that finished, it wouldn't be very hard at all to create a happybase package that just inverts the imports of the aiohappybase package to provide the strict supperset without code duplication that could serve as a nice transition period.

[aiudirog]

I've finished the "sync as a sub-package" implementation: https://github.com/aiudirog/aiohappybase/pull/3

It's a little hacky in its pursuit of code dryness but I think that's okay because in theory it will nearly guarantee the two implementations don't diverge without having to code everything twice.

[aiudirog]

I went through the open PRs and added counters to Batch and append() to Table. Between those two changes and some of the ones I had already made, #192, #175, #125, #96, and #89 should be satisfied.

[wbolster]

this is fantastic, thank you for all your work