aboutsummaryrefslogtreecommitdiff
path: root/contrib/netbsd-tests/fs/tmpfs/README
diff options
context:
space:
mode:
Diffstat (limited to 'contrib/netbsd-tests/fs/tmpfs/README')
-rw-r--r--contrib/netbsd-tests/fs/tmpfs/README17
1 files changed, 17 insertions, 0 deletions
diff --git a/contrib/netbsd-tests/fs/tmpfs/README b/contrib/netbsd-tests/fs/tmpfs/README
new file mode 100644
index 000000000000..cdc3f2d11ca7
--- /dev/null
+++ b/contrib/netbsd-tests/fs/tmpfs/README
@@ -0,0 +1,17 @@
+The tests in this directory where written at the same time tmpfs was
+developed. This is why, if you follow the order of tests in the Atffile,
+you will notice that they start checking the most basic things and end
+checking the less common ones. Furthermore, tests try not to use features
+tested by further tests in the lists.
+
+However, the above is not the most appropriate testing procedure when you
+have a working file system because some separation in test programs does
+not make sense afterwards.
+
+Many of the tests here are applicable to any file system. They should be
+refactored to be reusable on any mounted file system, which could also
+remove the need to do the mount/unmount steps in each and every test case.
+
+Possibly take a look at the file system tests in FreeBSD. They seem to be
+much more complete, even though they are written in Perl and therefore not
+directly usable.