Re: OO Progress Metrics?
From: Phlip (phlip_cpp_at_yahoo.com)
Date: 12/19/03
- Next message: Uncle Bob (Robert C. Martin): "Re: Why can't C# or Java support multiple inheritance? But inherit multiple interfaces instead??"
- Previous message: H. S. Lahman: "Re: class diagram question"
- In reply to: Uncle Bob (Robert C. Martin): "Re: OO Progress Metrics?"
- Next in thread: Randy A. Ynchausti: "Re: OO Progress Metrics?"
- Reply: Randy A. Ynchausti: "Re: OO Progress Metrics?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Fri, 19 Dec 2003 22:02:24 GMT
Uncle Bob (Robert C. Martin) wrote:
> "Randy A. Ynchausti" <randy_ynchausti@msn.com> might (or might not)
> have written this on (or about) Thu, 18 Dec 2003 19:15:41 -0700, :
>
> >It is just as easy to use an SLOC-based metric for estimating software
> >development manpower as it is using a story-based metric for estimating
> >software development effort. Why? Because each is a correlated
variable.
>
> The point I was trying to make before is that SLOC is *not* a
> correlated variable. Once again the bowling game is a case in point.
> There are implementations that take hundreds of lines of code, and
> implementations that take dozens.
>
> Another problem with SLOC is that it doesn't help us predict
> completion. We cannot reliably estimate the number of lines of code
> in a product.
>
> The story point scheme solves this problem by using relative estimates
> rather than absolute ones. It then *measures* the mapping between the
> relative estimates and real effort.
If an onsite customer (or equivalent) were to try to track a project's
velocity by plotting SLOC deltas for each feature, these would scatter all
over the place. Some 1-day features would add 300 lines, and some would edit
8 lines. The variable diverges, so tuning the process by that variable would
cause over-control. That means the results wander all over the place.
But when the onsite customer tracks estimated and actual times for features,
the variable narrows. The customer can tune the sizes of feature requests to
match the time periods programmers successfully estimate. Feedback will
cause results to converge instead of wander.
-- Phlip
- Next message: Uncle Bob (Robert C. Martin): "Re: Why can't C# or Java support multiple inheritance? But inherit multiple interfaces instead??"
- Previous message: H. S. Lahman: "Re: class diagram question"
- In reply to: Uncle Bob (Robert C. Martin): "Re: OO Progress Metrics?"
- Next in thread: Randy A. Ynchausti: "Re: OO Progress Metrics?"
- Reply: Randy A. Ynchausti: "Re: OO Progress Metrics?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
