[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: query
From: |
Russ Sherk |
Subject: |
Re: query |
Date: |
Fri, 17 Jun 2005 10:03:50 -0400 |
On 6/16/05, Jim Hyslop <address@hidden> wrote:
> Russ Sherk wrote:
> > On 6/16/05, Jim Hyslop <address@hidden> wrote:
> >>I'm curious - what use could this information possibly be, anyway?
> >
> > Usually this information is used by managers to determine churn.
> > Bigger churn (more files/lines changed) means bigger risk.
>
> Not if there's a proper set of unit tests in place.
>
> I'm always skeptical of raw numbers like this being used for any
> meaningful analysis.
>
> I don't think simply counting the number of lines added or removed is a
> good indication of risk. Suppose the tool reports "100 lines added, 100
> lines removed." Does that mean one line was changed 100 times? 100 lines
> were changed, one time each? Changing one line 100 times carries less
> risk than changing 100 lines once. And unless FishEye (or any other
> software) performs a fairly complex analysis of exactly which lines were
> added and removed, you won't know where on that spectrum your count of
> "100 lines added/removed" lies.
You are correct Jim. It should be used together with other metrics to
aid in determining the general health of a particular load or to see
general trends. E.g. we used to parse the logs between builds and
generate a list of changed files grouped by log entry and PR #. It
provided a snapshot of the changes that was easy to scan. (You could
see what changed, how much changed etc.) This was particularly useful
for T&V.
>
> --
> Jim
>
>
Regards,
--Russ
- query, Hridyesh Pant, 2005/06/16