On Mon, Mar 23, 2009 at 02:56:48PM +0100, Hans Aberg wrote: > On 23 Mar 2009, at 11:50, Cameron Horsburgh wrote: > >>> The problem, as I see it, is tied to the metric interpretation of >>> 4/4, >>> which is ambiguous: it can be taken as a strong beat (metric accent) >>> on 1 >>> followed by weaker on 2, 3 and 4. Alternatively, 3 can be accented >>> more >>> than 2 and 4, but less that 1: >>> | | | | >>> 4a > - - - >>> 4b > - (>) - >>> >>> Then 4a gets the beaming with the 1/4 note groups together and the >>> second gets the beaming with the 1/2 note groups together. >>> >> >> I'm not sure that's relevant---the behaviour happens regardless of >> whether it starts on beat one or beat three. >> >> The problem is that there is no way to tell the beamer to look forward >> an arbitrary number of notes when deciding whether to beam or not. In >> this case, the presence of a beam between the second and third notess >> in the group is entirely contingemt on the presence of a fourth *and >> fifth* note. If that fifth note exists, there shouldn't be a beam. >> >> This is what Trevor means: when the beamer realises there is a fifth >> note in the group (i.e. it's not four straight quavers) it needs to >> 'backtrack' and remove the beam between notes two and three. >> >> Or have I mnisunderstood you? > > See my other post. The normal way to beam is to select a meter, > including metric accents and subaccents, thus building a hierarchical > structures of subpatterns, and then let the beaming follow that. Though > traditionally, one may not adhere to that very strictly, and more than > one patten may be used, but perhaps not the same engraver in the same > piece and meter. > > Now, 4/4 is an exception to that in the higher number of such patterns > that can be used. If one then wants to automate what traditionally is > merely an ad hoc choice, then the problem is to lay down rules for that > so that a computer program can do it.
And there's the limitation--there's no way to automate this currently. Currently I have to do it all manually. Not a big job, but given this is a reasonably common exception LilyPond should provide a way to do it automatically. I've also just noticed another thing: the figure a8 a16 a16 a8 a16 a6 is rendered a8[ a16 a16] a8[ a16 a6] which is my preferred option. In fact, I would prefer everything to be broken to one beat groups *except* for the special case of four quavers which can appear as one group. It might be easier to write a rule for that and manually beam the four quaver figure which is far less common in my score. > > Hans > > > -- Cameron Horsburgh Blog: http://spiritcry.wordpress.com/ _______________________________________________ bug-lilypond mailing list bug-lilypond@gnu.org http://lists.gnu.org/mailman/listinfo/bug-lilypond
