X-Git-Url: https://git.cworth.org/git?p=apitrace-tests;a=blobdiff_plain;f=cli%2FREADME.markdown;fp=cli%2FREADME.markdown;h=4f49822a768006d33fdec56bd6448c1e681870ce;hp=0000000000000000000000000000000000000000;hb=5efeef24814e1744e3c4f3109f82e700576439ab;hpb=0ba9c4e22bc917a666b8762b734884e72c142ccf

diff --git a/cli/README.markdown b/cli/README.markdown
new file mode 100644
index 0000000..4f49822
--- /dev/null
+++ b/cli/README.markdown
@@ -0,0 +1,39 @@
+This directory tests high-level functionality of apitrace command-line
+interface (cli). See also the neighboring "traces" directory which
+also tests the cli, but in ways that can only be verified by doing
+comparisons of resulting dump files.
+
+For writing new tests, if the functionality can be exercised by a
+sequence of apitrace command invocations, and then a final comparison
+of "apitrace dump" output, then it's likely simplest to write a new
+test in the traces directory. Otherwise, a new test program can be
+written in this directory.
+
+The tests in this directory are found in files with names matching
+*.script by convention. The scripts must be listed explicitly in the
+CMakeLists.txt file. Each script consists of simple line-based
+commands with the following meanings (based on the first word of each
+line):
+
+  apitrace:     Execute the current apitrace executable being tested
+  		with the given arguments. If apitrace returns a
+  		non-zero status, the test will fail.
+
+  expect:       Compare the results of the previously-executed command
+                with the given (json-quoted) string. If the strings
+                are not identical, the test will fail.
+
+  rm_and_mkdir: Remove any existing directory of the given name and
+                then create it. The directory name is always
+                interpreted locally. If this fails for any reason
+                other than "file does not exist" the test will fail.
+
+Note: Blank lines and lines beginning with '#' are ignored.
+
+Commands can be prefixed with "EXPECT_FAILURE:" to indicate that a
+command is expected to return a non-zero value. In this case, a return
+value of zero from the command will cause the test to fail.
+
+If none of the commands in the script cause the test to fail (as
+described above), then the test will pass.
+