coil not same as protein nonhelix nonstrand?

[Elaine Meng]

If I open some protein (say 2gbp) and use these commands

cartoon style helix modeh tube radius 1.5
setattr coil residue is_strand true
cartoon style strand xsection round width 1 thick 1 arrows f

...there are still segments that are thinner than 1x1. Since I'd reassigned coil as strand, shouldn't "strand" get everything that's not helix? If I instead use ~helix it does get everything:

cartoon style ~helix xsection round width 1 thick 1 arrows f

[Elaine Meng]

the thin segments separate individual strands and helices; if the whole protein is changed to strand and ss_id for the whole protein is set to a single number, there are no thin segments.