Unit tests and benchmarks

[xxxserxxx]

The code base is really sparse on tests and benchmarks; this is a placeholder for that. I'll close this when code coverage is 50% or so. The widget code probably won't be testable, but as the devices are abstracted there's little reason why tests can't be added for the majority of the code.

Benchmarks are very important for this project, and there are none. Dependency choices have a large impact on the tool, and while I've done a fair amount of benchmarking, it's all been outside of the project. I don't have a target for benchmarking, but some will be better than none.

[AtakanColak]

@xxxserxxx hi, I had a question regarding this project and figured this issure thread would be a suitable place to ask it. In fact we actually discussed this a few months ago, but alas I still couldn't solve my issue. I'm currently using shirou/gopsutil to gather process information on linux. However I observe high cpu usage as much as %1000 on some test machines with high number of processes. I took a look at gotop and saw that even though gotop was also utilizing shirou/gopsutil for some cases, it didn't use gopsutil to gather process information; it used ps. Did you do any benchmarks between linux gopsutil and the current ps version?

[xxxserxxx]

Hi @AtakanColak . Yes, I did some benchmarking in preparation for replacing the ps code with gopsutil (for consistency), and also saw the same high CPU use. The ticket in gopsutil does have my benchmark code -- it's the comment that's one or two above your comment in the ticket. I haven't run it recently, and it's not yet integrated with gotop tests.

We moved a few weeks back, and it'll probably be a couple more weeks before we finish with the move and I can get back to the project, but let me know if you meant something other than the benchmarks on #842.