This patch makes the launcher script prefer `python`, when present, over
`python3`. This way, the launcher uses the user's preferred `python`
(which is often set up as a symbolic link to a particular python2.x or
python3.x), when available.
By default, if output is a TTY, dotbot will colorize the output. This
patch adds the option to pass `--no-color` to dotbot to have it suppress
this colorization.
Previously, PyYAML was hosted on BitBucket, so we had a mirror of the
repo on GitHub. Now, official hosting has moved to GitHub, so we can
point to the official repository instead. Thanks to Marco A. Feliu
<marco.feliu@nianet.org> for pointing this out.
This patch also updates the install shim to update submodule URLs. To
preserve the functionality of earlier Dotbot versions, we will need to
maintain 'https://github.com/anishathalye/pyyaml'. Because old versions
of the install shim used with new Dotbot versions will not update
submodule URLs, we will need to keep the old repository in sync with the
upstream repository as we upgrade PyYAML versions.
This patch also upgrades the dependency to PyYAML 3.12.
Prior to this patch, Dotbot was relying on running with the base
directory being the current working directory. In practice, it was
relying on the install shim to set up this context. It makes more sense
sense to actually execute `chdir()` within Dotbot itself, rather than
relying on the install shim to do so.
Because 'bin/dotbot' is a hybrid sh/Python script that finds the
appropriate Python binary and then runs, the test code should not invoke
it as an sh script when sh does not source the pyenv initialization
script.
The fix implemented in this patch is to directly run 'bin/dotbot' as a
Python script using 'python', which when called from bash, will run with
the correct Python version (because bash is set up to source the pyenv
initialization script).
This patch removes the snapshot/rollback system for the tests and
replaces it with a simple cleanup script that removes the appropriate
files from the VM's $HOME before every test. A side benefit is that
there is no more dependency on the Sahara plugin.
The cleanup script may not provide as good isolation between independent
tests as the previous setup, but it is orders of magnitude faster to run
without a VM rollback for every single test, and it does not break any
of the current tests.
