John wrote: | On Mon, 17 Dec 2001, James Allwright wrote: | | > My point is that missing out a start repeat is bad notation; an | > anacrusis at the start of a piece generates ambiguity and I think you | > will be hard pressed to find a music textbook that legitimizes the | > process of missing off start repeats. | | >From The Norton Manual of Music Notation, First Edition (Heussenstamm, | 1987): | | "If a passage is to be repeated from the beginning of a piece, only one | repeat sign is needed."
Yup; and there ain't a whole lot you're gonna do to fight this, unless you can somehow get control of all ABC software and add code to make it illegal. But on to a related, but new subject: A more seious problem is the common practice of omitting initial bar lines even when it's not the start of a repeat. This is another case where we can't fight it, but we could put subtle (or unsubtle) social pressure to change. I've run across this in the attempt to write code that does matching on the first few bars of a tune. The source of the problem is the question of whether the notes before the first bar line are a pickup or part of the tune. This is important, because pickups are notoriously variable. You want to exclude them from the match, because they will rarely match. What you want is to ignore them completely. But how does a piece of code recognize a pickup? The obvious answer is that a "pickup" is all the notes before the first bar line. But this doesn't work, because people often omit the first bar line when there's no pickup. You end up treating the first full bar as a pickup, which isn't what you want. So obviously, you count those notes, and if they're a full bar, you treat them as such? Not quite. It doesn't take much digging to learn that people are especially sloppy about their first bars, and the note lengths often don't add up right. You end up rejecting a lot of what should have been full bars because of this. Sometimes it's not even sloppiness; sometimes the first bar starts with a rest. Hardly anyone ever writes such rests, and the resulting first bar really does look like a pickup. This is a problem for live musicians, too, in some styles. At least it is to musicians who feel the difference between pickup notes and real melody notes. (And they'll likely get you in deeper, by pointing out that in some cases, the pickup is an important part of the tune which shouldn't be ignored. ;-) Another heuristic would be to say that an apparent pickup before the first bar is treated as melody if it's more than 1/2 of a measure. In the past couple weeks, I've transcribed several counterexamples to this. One was a tango, in 4/4 time, with 5/8 of a measure as pickup. This is not at all unusual in tangos. Another example was a tarantella, in 6/8 time, with 4/6 of a measure as a pickup. Tarantellas often have long pickups, sometimes 5/6 of the measure. So in both of these styles, even if the initial bunch of notes is only one tiny note short of a full measure, it might still be a pickup. Or it might be the first measure, which starts with a rest that was omitted. Or it could be incorrect note lengths due to the usual sloppy typing. So the obvious heuristics all have glaring counterexamples. What I've done so far is shrug and stick with my initial rule: Anything before the first bar line is a pickup, and is ignored. If someone doesn't write that first bar line, well, my code won't match their tune. The ideal solution would be for abc users to adopt the same policy. I can pretty much guess what are the chances of that ever happening. As I've noted before, we have a population of users who can't even be bothered to type "X:1" at the start of their tunes. And lots of printed music omits all initial bar lines, even for first measures. The publishers don't care whether this causes problems. Oh, well; pattern matches don't have to be perfect to be useful. To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
