Thanks to you Carl. Now, given that it's nonsense to offset an unknown value, I would remove this property from the offset command of Lilypond. Or at least, if that remove is not easily doable in the internal API, raise a *heavy* warning that you are using a random/deprecable command. Please do not see that as bad criticism. I'm telling that because it could save time to other people.
On Thu, Jan 16, 2020 at 12:22 AM Carl Sorensen <[email protected]> wrote: > > > > > *From: *Paolo Prete <[email protected]> > *Date: *Wednesday, January 15, 2020 at 4:16 PM > *To: *Carl Sorensen <[email protected]> > *Cc: *Lilypond-User Mailing List <[email protected]> > *Subject: *Re: Distance of a grob from its reference point > > > > I don't mean that with *broken*. I mean that it's unusable, given that the > values you put inside this function don't correspond to anything that you > can measure. Then, pretty random values. > > > > Please note that this doesn't happen with \override SomeGrob.X/Y-offset. > In that case, you can measure the offset with a ruler (in a very > uncomfortable way, though, given that you have to offset the ruler as well > with the ref point of the grob). > > > > Yes, this is true. Because when you \override you replace the > unpure-pure-container estimate function with a fixed constant value. > > > > \offset adds a fixed constant value to the existing result, which is an * > *estimate** rather than an actual value in the case of a Y-offset whose > default value is a unpure-pure-container function. The fact that you are > offsetting an estimate leads to random values, since the difference between > the estimate and the actual value is not predictable before completing the > spacing algorithm. > > > > Thanks, > > > > Carl > > > >
