[jzdziarski]

I don’t even understand what we’re comparing here. Is it lines of code? In that case, you must include all of the code from tr, sort, and any other shell commands used to perform the task at hand. By that standard, Knuth still wins by a long shot. Were this an actual coding contest (if such things exist anymore), one does not simply argue that instead of coding the solution in the given language, I’m going to just use five other people’s solutions and claim my trophy. Had the argument been who could most efficiently reuse all of the resources of an operating system to produce the laziest solution that will likely produce the highest number of obscure edge cases in the future, create unexpected dependencies, and likely scale the worst, then perhaps the shell script wins. But I think Knuth was trying to demonstrate quite the opposite - good coding praxis that can be maintainable, debuggable, and made to scale. I guess what I’m saying is that it’s difficult to see any comparison here, whatsoever, since the purposes were so markedly different. The shell script is not a solution any professional would ship in a product, and would only appeal to lazy one off tasks. Who would have been daft enough firstly, to not read Knuth’s paper, but secondly to think there is anything worth comparing in the first place? The script might be the quickest solution. Knuth’s code is the proper solution.