A current-limiting bench supply is an essential tool when prototyping. If one has enough space, and is operating at low enough power levels, that having the supply draw the maximum current from its source transform and dissipate all the power as heat would not pose a problem, a transformer, a regulator, some caps, a handful of resistors and two LM317 regulators would suffice [without current limiting one would only need a single LM317]. Commercial bench supplies aren't exactly free, but if your labor is worth much a small unit may be worth the money.
If you don't want to buy a commercial supply, it may still be worthwhile to build a simple board with a current limiter that has a few settings (perhaps use a switch instead of a pot if you like), a few fixed-voltage outputs (e.g. 5.0 volts and 3.3 volts), and a variable output. One could probably build such a thing on perfboard with an hour or two of work.
Even a board which only has an approximate 20mA current-limit setting may be useful. In many cases, one can program a processor-controlled board in such a way that it should take less than 15mA [by not having the processor enable any functions that would take more]. If one powers up such a board with a non-current-limited supply and something isn't right, it's possible that a mis-wired part could fail in such a way as to cause widespread damage. If one powers up the board with a roughly-20mA current-limited supply, the 20mA current limit will generally be low enough to prevent immediate damage.
PS--It's entirely possible that an accidental short which causes 20mA to flow where it shouldn't will cause some parts to be operated outside their Absolute Maximum Ratings. Any time a part is operated outside its AMR, one should expect that it may have been damaged thereby, and be aware of the possibility that a such damage might ironically alter a part's behavior in a way which would cause it to work in one's application even though an undamaged part would not. Nonetheless, when one's trying to get a design working for the first time, it's often helpful to assume that things will have probably escaped damage unless one sees evidence otherwise [especially when a 20mA current-limited supply makes it likely that they will have done so].