Hi Michael,

All looks good to me.  Some comments.

* (Perhaps) Controversial

    + The primary consumer of a finished spec. is QA

So, the language used in the spec should be oriented toward QA
personnel (i.e. less advertising or "selling" tone, but more
technicality and correctness).

        + UE - get involved before before the feature is
          implemented, or have some veto role

Not sure how this works out for a feature that arrives "complete",
that is, for a feature that was originally external to OO.o (hence
missed the initial discussion phase), but considered worth inclusion
at a later time.


* Proposed Changes

    + Unit Tests
        + we need a good, standard, very easy to use unit
          test framework, that can trivially be extended.
        + if something is easily unit-testable, it should
          have (at least some) unit tests developed in
          parallel to the implementation.

Some informal guidelines may be helpful for developing unit test cases
for a new feature once such framework is in place.

* Proposed New Process principles

    + Make simple things simple
        + a fast workflow for 'uncontroversial' changes
        + clarity wrt. scope and depth of communication

Bravo.  Simplicity is a virtue.

    + Split & document development practice guidelines:
        + HOWTO: iTeam formation (for those that want to)

Right.  This should not be mandatory.

        + split concept of 'design requirements' / development
          process, from the 'specification' [though there
          may be overlap]

I agree.  Splitting the specification into more specialized subsets
would be a good thing.

Also, I would like to see this list posted somewhere, and have us
revisit it in 6 months or so to measure progress.

Thank you all.

Kohei

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to