| PackageDescription | This module works with Test::More to create temporary directories that
stick around if tests fail.
It is loosely based on Test::TempDir, but with less complexity, greater
portability and zero non-core dependencies. (Capture::Tiny is
recommended for testing.)
The "tempdir" function is exported by default. When called, it
constructs a directory tree to hold temporary directories.
If the current directory is writable, the root for directories will be
./tmp. Otherwise, a File::Temp directory will be created wherever
temporary directories are stored for your system.
Every *.t file gets its own subdirectory under the root based on the
test filename, but with slashes and periods replaced with underscores.
For example, t/foo.t would get a test-file-specific subdirectory
./tmp/t_foo_t/. Directories created by "tempdir" get put in that
directory. This makes it very easy to find files later if tests fail.
The test-file-specific name is consistent from run-to-run. If an old
directory already exists, it will be removed.
When the test file exits, if all tests passed, then the
test-file-specific directory is recursively removed.
If a test failed and the root directory is ./tmp, the test-file-specific
directory sticks around for inspection. (But if the root is a File::Temp
directory, it is always discarded).
If nothing is left in ./tmp (i.e. no other test file failed), then ./tmp
is cleaned up as well.
This module attempts to avoid race conditions due to parallel testing.
In extreme cases, the test-file-specific subdirectory might be created
as a regular File::Temp directory rather than in ./tmp. In such a case,
a warning will be issued. |