I strongly suspect that most work was spent on hardware support for the different dive computers and maintaining the drivers. It would surprise me if Linus did not have a bunch of scripts that automatically filled in most of the log fields, which then explains why so relative little efforts has been put into the UI.
I would imagine that the pain of maintaining the hardware support for dive computers is similar to supporting the computer inside cars. Multiple manufactures, multiple models, basically no standard and everyone wanting to lock the users into their own ecosystem with their own cloud sharing social network, and likely filled with ads for products of their own stuff and paying partners. A open source project would basically have to choose a very small subset which happen to be similar enough to reverse engineer the protocol, a handful specific devices which are so extremely popular that there will be a developer dedicated to each of them, plus that one company which is open source/free software friendly.
If I recall right from one of the talks, subsurface has received/bought a lot of computers in order to make them work properly with subsurface so the relation between the different hardware, subsurface and libdivecomputer is as far as I understand it quite intermingled.
From the interview I get a mixed feeling where most effort was spent. A lot of the article focus on GUI and getting cross-platform working, but in the end linus remarks that "The biggest problem spot is probably hardware and particularly Bluetooth". For me this reflect my own views of subsurface and getting bluetooth to work with my own dive computer took a while.
I would imagine that the pain of maintaining the hardware support for dive computers is similar to supporting the computer inside cars. Multiple manufactures, multiple models, basically no standard and everyone wanting to lock the users into their own ecosystem with their own cloud sharing social network, and likely filled with ads for products of their own stuff and paying partners. A open source project would basically have to choose a very small subset which happen to be similar enough to reverse engineer the protocol, a handful specific devices which are so extremely popular that there will be a developer dedicated to each of them, plus that one company which is open source/free software friendly.