unity-dev team mailing list archive

[Ted Gould <ted@xxxxxxxxxx>]

On Mon, 2013-01-07 at 18:13 +0000, Mark Shuttleworth wrote:
> On 01/07/2013 03:21 PM, Ted Gould wrote:
> > 
> > I think that's the issue, you should be saying "We need to fix bug
> > 1234" not "Trunk was known broken, why aren't you focusing on my
> > problem RIGHT NOW."
> 
> 
> LEAN development treats a broken trunk as a stop-the-line problem,
> which is exactly "fix it RIGHT NOW". A broken trunk, or tests failing,
> is a stop-the-line issue at any shop that takes quality seriously.


To be clear, we're not talking about a failing trunk, we're talking
about failing integration testing (my fault in the example above).
Usually these are the types of failures that require investigation and
may not be directly related to the project at hand.  Serious, most
definitely.  But understanding the trade offs and time commitments is
also important for these types of failures.

Honestly, I think we're beyond the level of "trunk breaks" in our
quality processes.  That's for LEAN processes that allow developers to
commit to trunk.  I think we're more mature than that (or at least on
the cusp) for most of our projects.  Our bigger problem there is that we
don't have comprehensive enough unit tests to have reasonable faith that
we're not screwing integration further down the line.

Ted

Attachment: signature.asc
Description: This is a digitally signed message part