- Checked out twiki4 from svn
- Configured successfully
- pseudo-installed -link default
- chmodded everything g+w so that I can write to everything
- default plugins linked
- all files owned by apache user
- Apache 2, no mod_perl
Ran unit tests individually as myself with results:
- AccessControlTests, CoreTestSuite, FuncTests, HierarchicalWebs, PrefsTests, RegisterTests, RenameTests, SaveScriptTests, UsersTests, VariableTests
- repeated messages of the sort
Could not write "TWikiGuest | save | WebPreferences | | |" to /twiki/vanilla4.example.net/twiki/data/log200606.txt: Permission denied, Temporary web varying depending on test
Apache::Htpasswd::htDelete - User notauser not found in /tmp/junkpasswd: at /twiki/vanilla4. example.net/twiki/lib/TWiki/Users/ApacheHtpasswdUser.pm line 79
- RcsTests: repeated instances of
.../usr/bin/rcsdiff '-q' '-w' '-B' '-r1.1' '-r1.2' '-ko' '--unified=3' '/twiki/vanilla4.example.net/twiki/data/TestRcsWebTests/RcsDiffTest.txt,v' failed: Illegal seek
.test '1' '-eq' '2' failed: Illegal seek...sh '-c' 'echo urmf; exit 7' failed: Illegal seek..
TestRunner.pl: readline() on closed filehandle FH at /twiki/vanilla4.wildwoodweb.net/twiki/test/unit/TWikiTestCase.pm line 182.
The rest succeeded.
Last time I did this as superuser all succeeded, but that's not supposed to be necessary.
This was resolved on IRC. Don't know why Meredith didn't discard it.
No, actually, it wasn't
resolved. That's why i posted this. I am unable to get many of the unit tests to run properly unless i'm superuser, which I'm not supposed to have to be. I hoped that if I listed the specific problems I was having that someone might figure out what the problem is.
This leaves me unable to run most of the units tests, which I know you're trying to encourage people to do.
If the tests reach the end without reporting any failures, then the tests passed. Some tests are nosiy, becuase perl is noisy e.g. the illegal seek can be ignored. If you (the user) doesn't have permission to write to the log files, you will also see non-fatal noise.
Since you don't give any stacktraces above (a failing test will always provide a stacktrace) I have to assume all the tests pass - which is why I discarded this.