Open Stack

On 05/14/2013 09:37 PM, Daryl Walleck wrote:
>>From my experience, doing a plain text test plan for applications with the complexity of Nova doesn't scale well. I tried that, but without ways to intelligently sort/search/group test cases, it became unmanageable when I actually needed to pull data from it. I've been tinkering with a test management tool based on Google's ACC methodology (http://code.google.com/p/test-analytics/wiki/AccExplained) that's solved some of my issues with managing test cases. It's definitely not perfect, but I'd be open to sharing what I've worked on and how I've broken out my test cases.

Sure, interested to see your experience.

When someone mentioned test-analytics, my quick attempt kept me waiting at
"Loading project list".

Meanwhile, I recently came across ASCII doc -- -- http://asciidoc.org/, and
heard some very good reviews in community

Reviews comparing it w/ various light weight markup languages:

"I had the opportunity to seriously evaluate and contrast the serious
lightweight markup languages: reStructuredText (and its complement Sphinx),
Textile and AsciiDoc. Hands down, AsciiDoc is the winner. If there's such a
thing as a sure thing in software, AsciiDoc is it."

    - https://community.jboss.org/message/721016
    - https://plus.google.com/114112334290393746697/posts/CdXJt6hVn5A

Of-course, we don't *have* to take it on its face, but maybe doesn't hurt to
evaluate?

My 02 cents on why plain text ? (these are plain obvious, but just spelling
out):


Pros:

  (1) Text file are fast, easy to manage. I bet, they'll still be here 50 years
      later! These bulky "management systems" - I doubt that.

  (2) Works well over bad internet connections/remotely (I often work w/ a 2GB
      usb internet card). We can use tmux/screen session on a remote machine & work
      peacefully.

  (3) You can send patches, and apply them via the venerable git (and do all
      other magic).

  (4) Most (linux) developers are averse to clicky things. I've never seen
      someone review. However, they're (some) happy to contribute tests if they're
      text based approach somewhere in git files.

  (5) OpenStack uses rST (for all its README & docs & everywhere else).
      But, I'm aware, it's not the same as tests.

  (6) Add your own..


Cons:

  (1) Integration with Jenkins, etc ?
        - That's only needed for automation tests.

  (2) Reporting (although we could write scripts, but it's difficult).

  (3) Maintanability ?
        - git ?

  (4) I haven't seen any eamples of folks using text file based approach (that
      doesn't mean it shouldn't be explored.

  (5) .. ?


Don't hesitate to call me crazy, if I'm so. Thoughts ?

PS: Just to explore rST, I recently wrote some quick instructions while testing
nested virtualization w/ upstream kvm --
https://raw.github.com/kashyapc/nvmx-haswell/master/SETUP-nVMX.rst (not the
most beautifully formatted, but I don't hate it either):

Thanks.
-- 
/kashyap