Re: more config(8) patches (was Re: patch to fix breakage caused by earlier config(8) patch)

[David Cuthbert <dacut@xxxxxxxxx>]

ibotty wrote:
> why not [go overboard with code cleanups]? i understand, that when
> many  are working on a single point of code, these cleanups can be
> disturbing[...]

Having worked on such projects (that is, those which try to minimise the
amount of code changed), the rationale is that the more modifications
made to the code, the more likely you've introduced a bug.  Furthermore,
said projects have been burned by a case of someone performing code
cleanup, only to end up with a fatal bug shipping to a customer in what
was supposed to be a minor bugfix release.

Personally, I can sometimes agree with this for the so-called "dot" or
"point" releases (say, moving from 4.4.2 to 4.4.3).  However, I far too
often see a failure to branch this sort of code -- it ends up back in
the trunk (the codebase for 4.5 or 5.0), and more often than not
reflects an organisational fear of/apathy for testing.

Certainly, none of this applies to DF at this stage (as Matt has
indicated).

Dave