I'm not sure why this should be such a big deal. It was, in the end, as simple a thing as I thought it would be (although it sounded sort of scary: take a parallelization framework designed to run a program with different parameters on different machines, then aggregate the results, and make it transparently distribute calculation of a parallelizable function inside the middle of a Python script, while completely hiding the complexity of building a task list, taking your parameters from the command line, or even going through a separate program launch at all), and I implemented the first pass at it in exactly the time and space I thought: 200 lines (ignoring comments), in two days. I'm still surprised at how short it is, for what it does; Python is like that, though. (Incredibly expressive, at the cost of speed compared to non-dynamic languages.)
So now the team of computational biologists can just write calculation scripts that just do their thing and distribute them across our large overpowered number-crunching computer cluster without worrying about any of the complexity involved in actually distributing the jobs. Frankly, my co-worker's job of writing the parallelization platform is more impressive (although the Go language gives it a similar degree of "astonishing elegance" as the Python code I'm working with), but I have now made it (okay, without testing done, "am in the process of making it", it's got a week to go) truly convenient to use.
What gets me, though, is something that eludes me to describe. I'm describing what was produced, the objective and concrete things that are; in the end, though, this felt cool because of what it was like. With exceptions for "ack, brain is full, need break", the code just happened. There was a next line to write, I didn't get lost in it, my design worked and fit reasonably well with what I actually implemented, it just came together. And I think the reason why this is cool is that it's not an outlier. Not anymore, anyway. This is how I did some pretty awesome stuff in college, but it sort of fell away at Microsoft. At my current employer, it's what writing software is like- so when I get to actually do that, it tends to be this successful thing I get to take pride in, and I'm still not used to that yet- but the idea that "this is how it is" has made me start to feel like I might be good at what I do, and that's a new experience for me.
I've migrated to DreamWidth. The original post is at http://kistaro.dreamwidth.org/474793.html. View comments at http://kistaro.dreamwidth.org/474793.html#comments; go ahead and use OpenID to post your own, or you can comment here.