All of the prior remixes of Nick Montfort's _Taroko Gorge_ rewrote the text, while leaving Nick's code unchanged or almost so. I thought that was a shame. I also thought it was an opportunity! Since they all essentially consisted of word-lists plugged into a schema, I was able to remix them together on two axes at the same time:* Combining the word-lists of any two poems;* Mutating the stanza schema.I also took the opportunity to randomize the color schemes of the pages. (But not the font choices or the background imagery that some of the poems indulged in. Optima for everybody, I'm afraid.)Nick's original poem generates a constant ABBA-C pattern, with some extra B's thrown in. This page essentially invents a new pattern (for example A-, or BC-BA, or CCC, or so on) for each block. The code for the pattern is on the left, and the generated output is on the right.To answer the obvious question: Yes, this page really does execute the code that's displayed in the left column, and it really does generate the text in the right column.The most entertaining part of this project was inventing a way to mutate Nick's original code, while still having it *look* distinctly like Nick's original code. If you're not painfully familiar with that code - e.g, if you're neither Nick nor me - go to http://nickm.com/poems/taroko_gorge.html and select "View Source" in your browser, and look near the bottom. You'll recognize the form of the "do_line()" routine. Some of mine are simpler than his; some are more complex.The *complete* code for this page is of course much more involved than Nick's (because it has to *generate* Nick's code, plus a lot more). But I tried to stick to Nick's coding style wherever I could.The blocks strictly alternate between a single poem (with a mutated schema, but the original word lists) and a mix of two poems. I thought that would be the best lead-in for someone familiar with the original poems.A (perhaps interesting) result of my mixing is that some poems dominate others. Yoko Engorged has the longest word-lists, Fred and George has the shortest; so if they get mixed together, the randomizer selects many more Beatles references than Potter ones. I could have adjusted for this, but I didn't.
(Source: Author's notes in the source code of the work)
