ichiran VS tune

Check out ichi.moe. The word detection and splitting is quite good, and the backend is available on Github as ichiran. Unfortunately for most sane developers, the backend is written in Lisp.

Here's an example from my codebase which uses a lot of creative naming. There are "suffixes" (which I guess is a grammar term) but also "patches", "penalties", "synergies", "segfilters" and so on which are the terms I made up solely for this code.

You say batteries-included, I see a kitchen sink. Not everything needs huge libraries like ironclad loaded in, and every new dependency is a potential breakage in the future. I like to occasionally look at the dependencies list in my projects' .asd and see if I can get rid of something. For example I used cl-str only for its join function... And then I saw how it's implemented. I mean, really???. I rolled my own join instead. But if everyone starts using these battery-included kitchen sinks, I would still be loading a bunch of libraries I don't ever intend to use. I hear it's a big problem in node.js community.

I am working on Tune, an analytic/isolating conlang with each thing a separate word. I have only really worked out how I am going to treat nouns and verbs (the central classes of words), and "modifiers" (adjectives and adverbs, noun and verb modifiers). You append -i for verbs, -a for nouns, and-u` for features. You can have chains/phrases (like noun or verb phrases), they precede the head noun/verb and leave off suffixes.

I am thinking that by requiring phrasal verbs to be non-separable, they become more of a unit, and in the conlang I am working on at least, it would be easier to represent those than to handle the separable case. I am thinking will become "formal" verb phrases, which are like distinct standard concepts, as opposed to casual verb phrases, which might involve the same exact verbs but are made up on the spot.

I am working on a language called Tune for two reasons (still a ways to go):