You can usually commit a subtransaction to get a _p_jar and oid (I think) set. You can still abort this at the end of the test to clean up. hth, -Casey On Monday 28 July 2003 12:39 pm, Paul Winkler wrote:

I'm trying to write a functional test that verifies that an instance of my Product works correctly after it's been exported and re-imported. I'm stuck on how to give the test object a valid _p_oid which is needed to do the export.

Background: This Product is a dynamic content object for CMF. Its behavior depends on another CMF type instances, which it keeps a path to. So I need to ensure that things work even when one or both objects have moved. I've got moving & renaming handled well and now I want to test import/export.

So my test suite sets up a dummy CMF instance. I do this by swiping stuff from the CMFCore and CMFDefault tests; I subclass SecurityRequestTest and then do

manage_addCMFSite(self.root, 'cmf_test_site' ) self.site = self.root.cmf_test_site

I've also got a portal folder in the site which I keep a handy reference to as self.fol1.

All tests are passing, now I move on to writing the export / import tests.

Let's say the object I want to export is self.cmf_test_site.fol1.cp1. I have a shorthand reference to this stored as self.cp1. So I think the first thing I need to do is get the export data like this: data self.fol1.manage_exportObject('cp1', download=1)

... and once that works, I'll use the same data to do an import.

But the export fails. Using pdb and browsing some source I've learned that cp1 needs a _p_jar and _p_oid in order to be exported. The _p_jar I've already dealt with in my move / rename tests, by swiping a hack from CMFCore/tests/test_portalFolder.py:

id = self.cp1.getId()

# WAAAA! must get _p_jar set old, self.cp1._p_jar = self.cp1._p_jar, self.root._p_jar try: data = self.fol1.manage_exportObject(id, download=1) finally: self.cp1._p_jar = old

That seems to work fine.

But now I'm stuck on _p_oid. How do I give the object a valid one? I'm sure I can't just use the same trick and swipe the _p_oid of the app root :-P

I must say that using pyunit to write functional tests for zope 2 / CMF is a royal pain. It requires too much knowledge of implementation. (why should i need to worry about _p_foo???) Also I've found that if there's an error in your class' setUp(), it can cause the test to hang and never show you a traceback.

--

Paul Winkler http://www.slinkp.com

_______________________________________________ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )

... just a random remark: [...]

Also I've found that if there's an error in your class' setUp(), it can cause the test to hang and never show you a traceback.

I have run into a similar problem; this seems to be caused by the (usually reasonable) behaviour of the pyunit not to call the "tearDown" method of a test faiming in "setUp". Unfortunately in the context of unittesting Zope this results in a pending connection if the connection happends after the "get_transaction().begin()". After few tests, all availabe connections seems to be busy and the test hang. I have been able to fix this for my local tests by surrounding the code in the setUp with an 'try: ... excecpt:" block, where the "except:" block aborts the transaction before reraising the exception. This fixes the issue with the hang; one gets plain tracebacks telling what went wrong in the setUp. Looking at the more sophisticated structure of the CMF-tests this may be a bit tricky to apply this patch there. Cheers, Clemens