Whether you use a shorting or nonshorting switch, the net effect will be the same: currently unused caps will generate a "pop" when they are eventually placed in circuit. For that reason, I tend to recommend one of two other strategies: caps in series, or a blendable-bypass path.
When caps are placed in series, their combined capacitance can be calculated by 1/Ca+1/Cb+1/Cn = 1/C. So, .22uf in series with .22uf = .11uf. Place .47uf, .22uf, and .1uf in series with each other, and you get .06uf. It gets interesting when you shunt one of the caps with a straight-wire. Shunt .47 and you get .068uf. Shunt the .1uf instead, and you get .15uf. By use of a 3-position SPDT toggle (center-off) you can use a couple of caps and shunt either none, this one or that one, and achieve three different capcitances. The beauty of it is that all caps have a path to drain residual stored charge at all times, so you never get an audible pop the way you do with a rotary switch that leaves caps "hanging" and brings them into contact. Toggles take less space than rotaries, freeing up plenty of chassis surface for other things. The downside is that: a) you're limited to 3 choices, b) sometimes the math doesn't work out in your favour. of course, is you want more than 3 choices, you can always use a rotary to provide selective shunt ing of more than 3 caps.
Alternatively, you can use an approach Joe Gagan use productively on a number of his pedals, which involves having two parallel input caps, one of which is a much larger value than the other. The smaller one provides the default path for the signal, and the larger one is in series with a pot. As the pot resistance is decreased, more and more signal can efficiently pass via the larger cap, letting more bass in. Doesn't perfectly capture what selecting multiple caps does,but comes very close with less hassle and no popping.