Re: agile/xp question (formal analysis)

From: AndyW (foo__at_bar_no_email.com)
Date: 11/30/04 

Date: Tue, 30 Nov 2004 21:56:55 +1200

On Mon, 29 Nov 2004 21:21:47 -0500, Ron Jeffries <ronjeffries@acm.org>
wrote:

>On Tue, 30 Nov 2004 09:48:10 +1200, AndyW <foo_@bar_no_email.com>
>wrote:
>
>>>XP is certainly run in short iterations; but the time in those
>>>iterations is not broken down into phases as it would be in a mini
>>>waterfall. Requirements analysis, design, implementation, and testing
>>>are all done *concurrently* within the iteration. There are no
>>>milestones within an XP iteration that demark the completion of the
>>>requirements analysis for that iteration, or the completion of the
>>>design for that iteration.
>>>
>>Being you cant do analysis, design and imp concurrently, you need to
>>delay each stage at least a little to allow the output from one to be
>>used as the input to the other.
>
>Actually I can readily do design and implementation concurrently, or
>at least switch between them about every ten minutes. And I can
>implement features switching between analysis and design/imple every
>hour or half day. And the other members of my team can do the same
>thing, but not on the same cycle.
>>
Thats ok in a small team, scale it up and you wouldnt know enough
about the subject matter to do any of those things, design and code
yes, but not analysis.

Once you realise that you have to hand off to the subject experts the
methodology you are using needs to change. Scale it up far enough and
things get handed off enough so that coders become developers with a
valid opinion on the design process (usually implementing based on
rules given by the architecture team).

In a small team there are often combined roles that mean certain
techniques can be carried out efficiently. Scale those roles up and
you need to decouple them and provide an efficient communication
bridge. Thats where many small methods break down, because the
mechanism they use does not handle the assumed communication role
being made concrete.

Relevant Pages

  • Re: new here, my lang project...
    ... >> oo is better suited for the large scale, eg, more as an organizational ... functional languages, and some oo ones. ... As far as specific OOA/D methodologies are concerned, ... grand goals and "just can't get the design right". ...
    (comp.object)
  • Re: CPU power per watt. Can Forth help?
    ... Chuck left iTV. ... Design has to be in one hand as much as possible. ... Chuck's OKAD was originally a one-man show, ... and neither does it mean that it can't scale up. ...
    (comp.lang.forth)
  • Re: Determine size of tumbling block design
    ... You should be able to set the scale of the isometric graph paper to anything you want? ... Play about to get the design fixed, then draw it to scale - still on plain paper. ...
    (rec.crafts.textiles.quilting)
  • Re: Castle What?
    ... size of an A4 magazine rather than approx 70%. ... Obviously some items don't scale at all, ... image if you are using photocopy rather than printing techniques. ... and column width - and hence the design of the whole magazine. ...
    (comp.sys.acorn.misc)
  • Re: Determinism, -- a.k.a. physicalism --, has never been so alive and well.
    ... So for example, everything is energy, everything is "quality", even ... The idea that scale is fundamental is quite realistic. ... If you want to model the universe using cellular automata you will be ... If this is about design - everything can be ...
    (sci.physics)