Carl makes an even clearer case for what looks like the same sort of solution I spoke to earlier, by advocating a systematic distinction between paste operations that OVERWRITE, and those that REPLACE the hits in the selected target area. His analogy with text-processing seems to me to make sense here. Ben's point about "meaningless" time-signatures is a subtle one. ("Just gimme a 4/4 beat" is fine for lots of people, but any killer notation-based software package today has got to accommodate new-music composers and tala vidwans as well as jammers!) If, for example, you decide that two consecutive 4-beat bars should really be rewritten as 1 bar of 3 beats followed by one of 5, there's probably no easy way to do this now, i.e. to REBAR the grid leaving the hits unchanged. Pattern-based editing should make things like this much neater and simpler to do. Also, there are numerous ethnic-dance meters in which, for example, bars of four slow beats alternate with bars of three quick beats. In such cases, as Adrian suggests, things could still be handled just by setting different TEMPOS for the quick-beat bars, cumbersome though it may be to keep shifting tempos and display resolutions. But it's still nice to have the convention of /4 and /8 time-signatures to play with in expressing these relationships, in which the actual tempo really doesn't change at all. My hunch would be: Press on to complete the implementation of pattern-based editing, and then let's see what actual problems arise. The CMN-time-signatures-vs.-number-of-beats-per-bar problem is already half resolved anyway in DB, since resolutions are expressed only in number of subdivisions per beat, not in CMN-based fractions. And the grid itself is clearly beat-based. So maybe a wholly consistent labeling system isn't so important now. -Sterling