User Tim Gerrells ( posted:

Couple of thoughts:

1. This might be a time and place to simply create a manual test plan and
manually test through the current release. Automate what is high payoff
(core processing-related or user choke point UI functionality) and drive on
-- for now.

2. It is truly amazing how otherwise bright and intelligent and well-read
professionals deny objective reality and think that "it won't happen to
them" when deciding to risk abysmally poor software project management
characteristics, proven to cause the failure of 90+% of projects, only
hoping that they won't crap out when rolling those dice.

3. One alternative, albeit not a great one in reality, is simply state that
the required inputs to initiate alpha testing include: a) some form of
roughly complete functional requirements/specifications (I've accepted hand
written on legal paper, but made lotsa copies!); b) test environment as
needed has been configured and is ready (important when dependent upon
non-QA staff resources, as I often am); and c) successful first cut
"walk-though" smoke test to indicate the app is at least minimally stable to
begin testing. No meet-a da gate, no start-a da testing.

4. Normally takes serious kahonas and I've been told to move on to other
hunting grounds for one time I drew the line in the sand. But if the powers
that be act that way, you really don't want to work at such a place. It'll
soon show up on the rolls of former dot-coms or whatever, or you'll be
downsized, or blamed for failures you had nothing whatsoever to do with,
because there's an integrity and intestinal fortitude problem with upper

5. Did I mention, document the heck out of your conversations with