On 2018/06/17 10:54:38, dak wrote:
Right.  I haven't rebased this patch yet, but I naturally will have to
load up
another version for review now that the fix is in master.

I actually lean towards doing this in a manner similar to your original
approach, namely just using the stencil rotation call.  Basically I'd
want to completely replace both sten and s with a rotated copy.  This
will make a difference for later uses of s, notably

  // we use the bounding box if there are no boxes
  // FIXME: Rotation?
  if (!boxes.size () && !buildings.size ())
    boxes.push_back (Box (s->extent (X_AXIS), s->extent (Y_AXIS)));

The question is whether using the rotated version here is supposedly an
improvement, detrimental or "don't care".  Regarding the "relative"
implementation of the origin location, the bounding boxes resulting from
empty extents are likely to be garbage (likely NaN) after rotation.

https://codereview.appspot.com/344970043/

_______________________________________________
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel

Reply via email to