RE: [ft-devel] issue with ArialMT in 2.1.9

[George Williams]

Werner points out correctly that the T1 spec says that the first path
construction operator should be an rmoveto (in the section on hsbw)

However, the example glyph in section 6.6 says:
        "The initial rmoveto command (or its equivalent) is required..."
and then proceeds to optimize the rmoveto into an hmoveto.

My guess is that Adobe's intent is that a glyph must begin with one of
the *moveto operators rather than rmoveto alone.