Redo fossil segment to use RepoStats

[emansije]

Includes a hack to make it work in older versions of fossil that throw an error with fossil changes --differ, like version 1.37 does.

[b-ryan]

This looks great. I'm going to try to look in more detail & do some of my own testing today. I am a bit busy though so if I don't get back to you by Monday please feel free to remind me.

[emansije]

In the meanwhile I found a bug that manifested when the fossil repository had no changes in the tracked files but had untracked files. That situation would brake the parsing because it would introduce an unexpected blank element in the beginning of the status list.

[b-ryan]

@emansije how would you feel about introducing tests for this? I think the most helpful thing you could do would be to test the parse_fossil_stats test - you could use the uptime_test file as an example of how this might look.

[emansije]

Sure! I really should have implemented the tests from the start, it'd probably have saved me some time. I'll use your suggestions.

[emansije]

I can't get the test to work when the fossil segment is a threaded one, it keeps throwing this error:

RuntimeError: threads can only be started once

[b-ryan]

I just played around a bit to figure out what was going on. I made some changes here: ebb7187

The issue was the use of self.segment.start() within a for loop in your test_all function. The segment is created in the setup function but then started multiple times within the loop. The easiest way I could figure to get around this was to just test the parsing function rather than testing the segment itself.

Let me know what you think / if you have other suggestions.

[emansije]

Looks good to me! The tests and the segment work just fine and it's a cleaner solution, too.