Question about log simplification via radcan

>>>>> "Karl-Dieter" == Karl-Dieter Crisman <kcrisman at gmail.com> writes:

    Karl-Dieter> Hi! ?Here is a question
    Karl-Dieter> from?http://sourceforge.net/p/maxima/bugs/2341/
    Karl-Dieter> (thanks, Mike O.!) which hadn't gotten attention yet,
    Karl-Dieter> but I'm just curious about, so I'm posting it. ?Is
    Karl-Dieter> this a "wontfix" because of radcan's behavior as
    Karl-Dieter> symbols, not functions, or is there a real bug here? 
    Karl-Dieter> ?Obviously output 6 is log(2) for some values of log,
    Karl-Dieter> but that's not what I'm getting at; presumably log(2)
    Karl-Dieter> here means the usual thing. ?Thanks!

    Karl-Dieter> (%i1) display2d: false;

    Karl-Dieter> (%o1) false
    Karl-Dieter> (%i2) domain: complex;

    Karl-Dieter> (%o2) complex
    Karl-Dieter> (%i3) declare(t, real);

    Karl-Dieter> (%o3) done
    Karl-Dieter> (%i4) f(t) := log(2*t) + log(1/t);

    Karl-Dieter> (%o4) f(t):=log(1/t)+log(2*t)
    Karl-Dieter> (%i5) radcan(f(t));

There's something else going on too.

(%i1) logexpand;
(%o1)                                true
(%i2) logexpand:false;
(%o2)                                false
(%i3) log(1/t);
(%o3)                              - log(t)

I would have expected that logexpand:false would prevent changing
log(1/t) to -log(t), which is obviously wrong when t = -1.

Ray