This is an automatically generated e-mail. To reply, visit: http://reviewboard-openlmi.rhcloud.com/r/1147/

On October 30th, 2013, 2:44 p.m. UTC, Radek Novacek wrote:

src/software/test/testSoftware.py (Diff revision 1)
255
        self.assertEqual(sorted(package_list), sorted(rpm_packages))
use "assertListEqual", it produces better output when the test fails

On October 30th, 2013, 3:29 p.m. UTC, Michal Minar wrote:

Unfortunately this is available since python 2.7.

On November 5th, 2013, 12:09 p.m. UTC, Jan Grec wrote:

I'm afraid there are more pieces of tests code that won't work on 2.6

Shall we strictly support python 2.6 and more, or is it ok to run with 2.7 and eventually rewrite tests for older python versions when it's needed?
I'd rather stick with code compatible with 2.6. True, not all of the current code works with 2.6 but this is not the reason to use new features from 2.7 since we know these tests need to pass on 2.6 sooner or later.

On October 30th, 2013, 2:44 p.m. UTC, Radek Novacek wrote:

src/software/test/testSoftware.py (Diff revision 1)
284
The matching against 'yum list available' is missing. If it is on purpose, add at least a TODO comment.

On October 30th, 2013, 3:29 p.m. UTC, Michal Minar wrote:

This should be achievable. Try to remove installed packages and duplicities from lmi list and compare them. (I haven't tried).

On November 6th, 2013, 11:54 a.m. UTC, Jan Grec wrote:

Skipping this one for now.

Although it'd be great to know exactly whether listing available packages returns correct available packages, there's no exact process to get them in lmi itself.
>From packages returned, we'd have to remove all packages with package names that are already installed and their epoch or version is lesser omitting updates which requires package name parsing etc. etc.
I suppose this is going to be an openlmi-scripts functionality and therefore should be tested there, not in lmi itself.
Fair enough.

On October 30th, 2013, 2:44 p.m. UTC, Radek Novacek wrote:

src/software/test/testSoftware.py (Diff revision 1)
312
        package_name = self.cim.LMI_InstalledSoftwareIdentity.first_instance().path["InstalledSoftware"]["InstanceID"]
You can use

.InstalledSoftware.InstanceID

instead of

.path["InstalledSoftware"]["InstanceID"]

On October 30th, 2013, 3:29 p.m. UTC, Michal Minar wrote:

From what I know:
  .InstalledSoftware["InstanceID"]
Because InstalledSoftware is an instance of pywbem.CIMInstanceName, which doesn't allow for smart attribute access.

On November 5th, 2013, 12:09 p.m. UTC, Jan Grec wrote:

This shortcut somewhere works for me, but in this specific place is doesn't.

TypeError: 'LMIInstanceName' object has no attribute '__getitem__'
Please update your openlmi-tools (to version 0.9).
Actually now it's not "You can use", but "You must use".

- Michal


On October 30th, 2013, 2:03 p.m. UTC, Jan Grec wrote:

Review request for OpenLMI Developers.
By Jan Grec.

Updated Oct. 30, 2013, 2:03 p.m.

Repository: openlmi-providers

Description

software - tests: software test suite with basic tests

This is only a "current-status" review. I'll be very thankful for any comments on what's wrong, how the structure should look like, etc.

My TODO is to
  * repair all broken tests
  * add all tests mentioned
  * bind test suite with new LmiTestCase class

I apologize for inconvenience, but I wasn't able to save previous review notes.

Diffs

  • src/software/test/lmi-test.repo (PRE-CREATION)
  • src/software/test/testSoftware.py (PRE-CREATION)

View Diff