Currently, mkdir always succeeds for existing files and it
even modifies their mode. With this change, only mkdir -p for
existing directories will be allowed.
Fixes#1017.
test_mkdir_dup_dir asserted that creating an existing directory is an
error, but that's not how GNU coreutils behaves. This has been reported
in #121, but wasn't fixed (only the `-p` case was).
Although for some tests this adds characters
we still use them there because the
brevity cost is now worth the benefit in
terms of instant, natural-language readability
and recognizability for people not familiar
with this tests of this module or even the project
Updates to individual integration tests
- use proposed conventional approach to beginning tests
- use new convenience functions for using fixtures
- use new names for TestScenario
Updates to integration test modules
- add proposed conventional module-level functions
Updates to test/common/util.rs
- rename TestSet, and its methods, for semantic clarity
- create convenience functions for use of fixtures
- delete convenience functions obsoleted by new conventions
The main motivation is to move toward running those tests for a specific
target, that is, if a test won't run on Windows, then we shouldn't build
it. This was previously the default behavior and prevented a successful
run on AppVeyor.
I borrowed this pattern from the tests in the Cargo project.
