About the usage of test.error()/test.fail()/test.cancel()

[dzhengfy]

Recently we had some discussion to synchronize the same understanding for the usage of above functions in the subject.
Below is the current understanding we have now. Welcome to comment!

• The basic strategy is that we should skip tests as little as we can

• test.fail() - used only when failures happens for case checkpoints

• test.error() - used when

  1. vm starts failed
  2. create disk failed
  3. required package missing
  4. not enough vms
  5. incorrect parameters
  6. and any failure for test environment setup

• test.cancel() - used when

1. The feature does not support with current libvirt version or architecture
2. The hardware configuration does not support the test

[dzhengfy]

@sathnaga Welcome your opinions :D