<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/"><channel><title>Scott Graham's web log</title><link>http://h4ck3r.net/blog.xml</link><language>en-us</language>

<item><title>5 Aug 2007 &ndash; 12:09AM</title><description>

<p> I love having a hard piece of code to write. There's a vague sense
    of apprehension that I currently have no idea how the answer will
    present itself, so I semi-avoid the problem for a bit. I try to
    figure out a nice simple input for the problem, or maybe try to
    create a few decent examples of it. Depending on how hard it is, it
    might be time to check reddit after that, or maybe get a drink. At
    some point, I figure it's time to have at it.  </p>

<p> I normally start by writing things &quot;around&quot; the problem,
    where they're not really involved in solving the real problem, but
    they continually pare down the input to get just the core of the
    problem that's trying to be solved. And then, when the core problem
    can't be evaded any longer, it's time for a change in direction.
    Instead of writing the things that focus and pare down the problem,
    I start sketching some sort of half-assed solution to the problem,
    by using lots of magic functions, for the time being assuming that I
    knew a smart guy who could write the magic functions for me.  </p>

<p> By going back and forth between the two attack methods, dividing
    things up into little bits, at some point things just <em> click
    </em> (normally all of a sudden), and all that remains is to write a
    bunch of functions. The difference with these functions is that the
    only difficult part is being able to type quickly enough for your
    brain not to get bored.  </p>

<p> The answer has presented itself.  </p>

</description></item>

</channel></rss>
<!-- vim: set noai nosi expandtab indentexpr= tw=72: -->
