I think that few people really NEED true RMS capability for most run-of-the-mill measurements. We're usually working with sine waves, so the average-measure, rms-display meter works for 99% of the stuff you need. You know, we've gotten along pretty well in the years prior to the commonly-available true-rms meters, haven't we?
Whenever you have complex waveforms to measure, you're usually using an oscilloscope and looking for a lot of different characteristics that you can't see on a DMM, such as risetime, fall time, glitches, sag, pulse width, period, etc., especially on an incredibly-complex waveform such as a composite television signal.
I don't much like the bargraph on a DMM simply because it has stepped resolution. Each bar turns on only after you're reached the next step in the bargraphs range and that step may be a lot larger than you can tolerate if making fine peak or nulling measurements. A good example would be that of using a DMM with a good ACV frequency response to null out a homemade THD analyzer. You'd never see the best null with the DMM and that could affect the final measurement tremendously. If you need an analog meter's capability, use an analog meter!
Dean