Extending on the experimental pipeline, made a couple of changes today.
Cleaned up the entire code base and organised to proper header/sources.
Streamlined existing git helpers to start with proper remotes and create releases.
Now that the structure is set up, it was also time to drop the use of <algorithm> and std::sort. I had used a sort since the dev version for two uses - verify functionality and share. Given things are stable, decided the client can handle this.
As it stands, even the client is rather thin in that it takes the JSON and inserts to Mongo. Since the fields that I use downstream to sort are already present, the API can use the existing DB sort.
A funny side effect with splitting things into their own headers/sources is I've saved 4k. Not that it adds up much, but it gives some perspective into what happens as a whole. Oh! there's this interesting issue with std::mutex that I encountered too. Guess that's another avenue to go digging down.