Bug: simple chord inversions take 2-octaves

[Nucleos]

Hello,

c major is : <c e g>

The first inversion would be <e, g, c>. This is what I call c/e.
The second inversion would be <g, c e>. This is what I call c/g

The code below triggers nevertheless Lilypond 2.14.2 to display <e, c g>
-- which is 2 octaves wide !

Weirdly, Lilypond seems to think my definition of inversions is right
for c/g (it displays <g, c e>), but uses a weird definition for c/e. Why
is that?

The code :
********************

\version "2.14.2"

{

\chordmode {

c1 c/e c/g

}

}
********************


I suggest the behaviour should be:
1) build the chord "c" is <c e g>
2) when the bass of the chord is given, take this bass and then follow
the natural order of "c e g c e g c e g"... and display three successive
notes of that sequence.


Cheers,

-- 
Nucleos