4 ms·
Is anyone else puzzled by the idea of f.lux consuming 14 developer years (at least according to the article), it's an awesome app but that seems like a huge num
by forgotpwtomain 11y ago
Is anyone else puzzled by the idea of f.lux consuming 14 developer years (at least according to the article), it's an awesome app but that seems like a huge number for what it is: a gui interface to a very simple function*.
[0] http://www.tedunangst.com/flak/post/sct-set-color-temperature http://www.tedunangst.com/flak/post/sct-set-color-temperatur...
- joshu 11y agothere is a whole bunch of stuff built but unreleased outside of the app.
- kobayashi 11y agoWhat kind of stuff?
- joshu 11y agoIt is complicated. I only know because I am a friend, so we have to wait till he releases stuff.
- edwintorok 11y agoWill it be released as open-source? Doesn't really make sense to develop it as a closed-source application, when the application itself is offered for free, and open source versions exist already.
- verusfossa 11y agoIf this is true, it's unfortunate. Things like flux should have a small footprint and a small working domain. I'm willing to spare some complexity for usability tools, but always end up getting off the bus when it gets too heavy.
- keithpeter 11y agoQuote from interesting page linked in parent post... "...Now to be fair most of this is the result of OpenBSD ports choosing to build and package optional dependencies, including the GUI..." I suspect the people who produce and maintain the ports builds are putting all the options in to cover as many use-cases as possible. I've noticed that some OpenBSD ports give alternatives to pick from with different compilation options. I suspect that might be impractical for many choices. Now I'm wondering if I can build sct.c on the Slackware box and have a play with a bash script setting the colour temp by time of day for current latitude...