aboutsummaryrefslogtreecommitdiff
path: root/tests/test/script
diff options
context:
space:
mode:
authorBoris Kolpackov <boris@codesynthesis.com>2019-06-24 12:01:19 +0200
committerKaren Arutyunov <karen@codesynthesis.com>2019-07-01 18:13:55 +0300
commit977d07a3ae47ef204665d1eda2d642e5064724f3 (patch)
tree525a3d6421f61ce789b690191d3c30fc09be3517 /tests/test/script
parent7161b24963dd9da4d218f92c736b77c35c328a2d (diff)
Split build system into library and driver
Diffstat (limited to 'tests/test/script')
-rw-r--r--tests/test/script/builtin/cp.testscript13
1 files changed, 13 insertions, 0 deletions
diff --git a/tests/test/script/builtin/cp.testscript b/tests/test/script/builtin/cp.testscript
index 5f69f05..aa16465 100644
--- a/tests/test/script/builtin/cp.testscript
+++ b/tests/test/script/builtin/cp.testscript
@@ -358,6 +358,18 @@ if ($cxx.target.class != 'windows')
EOE
}
+ # @@ Temporary disable the test as it may fail if the package is fetched
+ # from the git repository location. Note that the used `ls -l` command
+ # displays the filesystem entry modification time with the minute
+ # resolution and the build2 dist meta-operation doesn't preserve the
+ # directory modification time. That's why the test command may
+ # unexpectedly succeed.
+ #
+ # One way to fix it is to instruct ls to show time with at least the
+ # second resolution. There is no portable way for that across different
+ # OSes but all of them seems to provide some option to achieve that.
+ #
+#\
: dir
:
{
@@ -372,5 +384,6 @@ if ($cxx.target.class != 'windows')
%.+
EOE
}
+#\
}
}