Qucik poll for those of you who have deployed #iocaine, or are considering doing so. Currently, I provide pre-compiled binaries of it, a single binary wrapped in various packaging boilerplate. That single binary includes a default script, a default word list, and some text to use for markov chain training embedded in it (and a default configuration too).
This makes deployment reasonably simple: it's a single binary. Drop it, run it, pretty much done. I don't think anyone runs it like that, however.
I'm considering dropping the embedding, and shipping not a single binary, but a tarball instead. A script, some configuration, training data and everything else that was embedded was still in the tarball, and it would be set up so that iocaine can easily find them - they wouldn't be embedded into the binary itself, is all.
Not embedding allows me to drop multiple dependencies, and allows me to remove a whole lot of code that made embedding work.
My question is: how would this affect you? Also feel free to reply with details, if you have some to share.
@algernon i'm using the docker version cause i'm not smart enough to figure out the nixos flake thingy.
i appreciate the ability to feed it different books like olde timey english like Shakespeare and such. wouldn't it be better if we all fed it different slop back? not the same ones that come precompiled with it?
long as there is a docker, do what's easy for you though.