Open Stack

> On Jun 23, 2015, at 9:46 PM, Rob Hirschfeld <rob at zehicle.com> wrote:
> 
> All,
> 
> I was looking at the capabilities subdivision lists [1] to see if I could get that done before the meeting tomorrow (which I will miss).
> 
> In reviewing the first item, I found that the spreadsheet has a lot more tests than we currently are tracking in the guideline.
> 
> For example, "compute-auth" has 4 tests currently, but the 10 new divisions [2] have 28.
> 
> That presents some items for consideration
> 1. we have a lot more tests coming that will impact users of the guideline (RE potential flags)

Yup. We can get a jump by running against existing clusters and seeing what falls out.

> 2. it will require automation to make the change effectively (as Chris Hoge pointed out)

Review in progress. I included my hacky python script which tries to reconcile the current .next
and proposed reorganization in a reasonable way, but to also encourage others to make
suggestions on how to do it better. There may need to be some manual tweaking done,
especially to the descriptions of the capabilities (which are lifted from the original ‘next’ file).

> 3. I don't know how to track which tests we have chosen to omit based on D302

I would suggest a file of excluded tests, or maybe even a new category in the json file (which
should be backwards compatible since we’re adding rather than taking away from the 
dictionary).

> 4. we'll need people to review this pretty carefully!

Please look at both the json file and the code. Errors in the json will be because of bad
code.

> Even if we don't subdivide the capabilities, we will need to bring in these tests.

Let’s do that as a separate patch. I feel like the first step is to reorganize the existing
file, the build upon that.

<rest snipped>