Re: [xml-dev] MicroXML and &

[Rick Jelliffe <rjelliffe@allette.com.au>]

On the other hand, the kind of complexity that extra lines in a table
cause is trivial, perhaps negligible after the first entry.   (You
wouldn't implement a  large lookup table as an if-then-else chain,
surely.)

I think it is only sound to use the Source Lines of Code metric to
measure code complexity when it is, for example, a reasonably proxy
for Decision Point Analysis or some other complexity with risk
attached: in the case of a lookup table (not used for jump vectors),
SLOC clearly isn't a reasonable proxy.   (It might be consideration if
you had to type numbers in by hand, due to the possibility of
transcription error, but I don't expect your would be doing that
either.)

The line argument is a little strange to me: if the absolute number of
lines is the problem rather than complexity (because you only have a a
certain number of sheets of paper in your printer? because you are
programming on an android phone screen?), you could just have multiple
entries per line.  Reduced to the absurd,  I guess it means we could
simplify the whole program by over 4000 times by putting everything on
one line?  :-)

Cheers
Rick






On Fri, Jun 1, 2012 at 6:15 AM, John Cowan <cowan@mercury.ccil.org> wrote:

> There are 2236 entities in the current draft of the W3C entity set, but
> only 2117 lines of Java in my parser.  I don't think I want to double
> its size.