Jonathan Nieder wrote: > Stefan Beller wrote: >>> Stefan Beller wrote: >>>> Nowadays there are better tutorials out there such as "Git from bottom up" >>>> or others, easily found online. Additionally to that a tutorial in our >>>> test suite is not as easy to discover as e.g. online tutorials. [...] >> 2ae6c70674 (Adapt tutorial to cygwin and add test case, 2005-10-13) >> seemed to imply that it was testing some part for Documentation/tutorial.txt >> though. > > Oh, good point. > > v1.2.0~121 (New tutorial, 2006-01-22) means that the test is no longer > testing what is in the tutorial in any meaningful sense. That's why > my search for "git whatchanged -p --root" in manpages didn't find > anything. Correction: the tutorial is now called gitcore-tutorial and mostly survives. A search for -p --root failed because of v1.5.5.1~19^2 (core-tutorial.txt: Fix showing the current behaviour, 2008-04-10). That said, the conclusion that this test has mostly bitrotted as far as its original purpose goes still applies. An alternative method of addressing the goal you described would be to fuzz object-id shaped things out of the sample output. I don't have a strong preference, given how little this test contributes to coverage (as you mentioned) and how difficult it is to make it continue to match the documentation it is trying to test. Thanks and sorry for the confusion, Jonathan
