Oh good grief this is big, I'm really sorry about it. A bit of the history, when I first started _ng I did add a few things to keep up compatibility w/ newer Erlang and that kind of grew and grew to more things including AAE, dialyzer updates and quite a few cherry picks.
Now the own fork was kind of Ok for me but if we want to make this usable it seems sensible to go back using 'develop' as the central branch.
The code does compile, run eunit and eqc tests w/o failure but I'm pretty sure I've messed some things up so I don't want to blindly merge this in.
EDIT:
I forgot it also switches from rebar2 to rebar3 and from git to hex dependencies.
Sorry about the late reply, the mail swamp got me.
The most notable changes are:
adding AAE form riak_kv.
R19 compatibility with a special _rand module as there some tricky corner cases as core depends on some rather specific behaviour of the old RNG.
making it all use hex
Fixing dialyzer specs for R19 dialyzer (mostly missing undefineds)
Some improved error reporting
extracting chash as an own re-usable library
adding some -callbacks that were missing
updating the worker pool to allow both fifo and filo message handling
Oh good grief this is big, I'm really sorry about it. A bit of the history, when I first started _ng I did add a few things to keep up compatibility w/ newer Erlang and that kind of grew and grew to more things including AAE, dialyzer updates and quite a few cherry picks.
Now the own fork was kind of Ok for me but if we want to make this usable it seems sensible to go back using 'develop' as the central branch.
The code does compile, run eunit and eqc tests w/o failure but I'm pretty sure I've messed some things up so I don't want to blindly merge this in.
EDIT:
I forgot it also switches from rebar2 to rebar3 and from git to hex dependencies.
Sorry about the late reply, the mail swamp got me.
The most notable changes are:
_randmodule as there some tricky corner cases as core depends on some rather specific behaviour of the old RNG.undefineds)