The developer's hell - do what's necessary or do what's fun.
My approach at work is to devote a bit of time throughout each week to something that is fun – something that's not immediately useful, but which can come in very handy somewhere down the road.
I tend to deal with a ton of high-pressure stuff all day long and some of it is detailed, fiddly and in which it is easy to miss something (such as revving multiple iterations of complex Hype projects). When it gets to be too intense, I do something fun instead - such as a Hype project that some something simple but very cool (such as the interlooping swirls I posted a few months back).
Many times over the past few years I've been able to pull one of these out of my hat, pop in into place and suddenly I'm a miracle-worker. (Wow! You did that in only three minutes?!!!) Of course, it took weeks or months of five minutes spent here and there, but the miracle worker thing is definitely good for the rep.
If it is low priority but fun, set aside a half hour a week for it. When it is finally ready and introduced out of the blue, people will be very impressed.