Something else that's been bothering me, overtone uses a mix of println and java logging. I don't think using a real logging framework makes much sense here. There's some benefit in being able to up the verbosity, but because Java logging it's such a mess it's also unpredictable. I noticed for instance that in minibeast the logs don't show up. I don't want to help someone debug their logging setup first before I can actually get some debug info... It also writes a log file under .overtone. has anyone ever looked at that? When would that be useful?
You don't think it would be worth using timbre or something similar?
I dislike timbre with a passion.
Then I'd rather use a standard solution based on slf4j/logback.
Why so? I'm not exactly a fan, but it does the job and seems to be the default choice for clojure projects.
I would hardly call it the default, although we do see it regularly on client projects. For Gaiwan/Lambdaisland projects we default to pedestal.log/logback and glogi. Plenty of people also use clojure.tools.logging, and mulog is gaining ground and is probably more the direction we should be heading in. The problem with timbre is that it muddies the water by trying to reinvent the wheel. The java logging landscape is a mess (as I've written about in the past), but at this point it's a well understood mess with a few common reliable configurations. Timbre adds another layer on top of that, which reimplements a lot of the functionality that the lower levels already implement, but breaking common expectations. Eg loggers in timbre aren't hierarchical.
Thanks. Just checking out mulog now. Looks interesting.