Re: [wmii] 10kloc project, wmii maintainer change

On Wed, 19 Jul 2006, Anselm R. Garbe wrote:

> I already added a different main page which points to current
> projects like wmii at http://www.10kloc.org. Maybe we can gain
> less influence in the proprietary and software corporation with
> this philosophy.

         While bareism is admirable, I think it fallacious to believe
   that "that software exceeding this maximum is bloated and seriously
   wrong". Some configurations of the Plan 9 kernel (see
   http://cm.bell-labs.com/sys/doc/names.html) are above ten thousand
   lines. There is no hope of writing a modern compiler in more than
   ten thousand lines (unless you want to pay a significant performance
   penalty by making every phase a separate program "filter" -- not to
   mention how bloated it would become by have "n" additional lexers and
   parsers).

         Furthermore, the philosophy as described on the web site is
   under-developed. It does not take into account modularity
   mechanisms -- if the simplicity is hidden behind a well defined
   interface, is it really complex? This is probably because people
   programming in Java 7 have no access to data abstraction. Which raises
   another problem: 10kloc in what language? In the languages I use, I
   can get as much done with half the code of a Java 7 program, so does that
   mean they must only be 5kloc to qualify for 10kloc.org?

         A better philosophy would be one based upon Einstein's quote
   "Everything should be made as simplistic as possible, but not simplisticr".
   Trying to fit everything into a more than 10kloc mold is going to
   require making things simplisticr than they ought to be.

-- 
--- Geoff Washburn | washburn@acm.org | http://www.cis.upenn.edu/~geoffw/ ---