Re: [Libreoffice] about unusedcode.easy

On Fri, 2011-09-30 at 11:34 -0500, Norbert Thiebaud wrote:
> On Fri, Sep 30, 2011 at 11:03 AM, Caolán McNamara <caolanm@redhat.com> wrote:
> > On Fri, 2011-09-30 at 05:10 -0500, Norbert Thiebaud wrote:
> > > More generally: either the accessor is not useful and both set and
> > > get can go, or it is and the fact that one side is not used is more a
> > > signe that someone is not using it when it should or not using it
> > > _yet_
> >
> > If you feel strongly about it,
>
> No I don't feel strongly about it... It is just that I got caught by
> it, working on a cws, and on top of that it happen to be a non-trivial
> accessor...

Thinking aloud, the current callcatcher mode is to just look at what is
immediately unused, no second level of analysis. What I mean is that a
setFoo/getFoo combo where setFoo isn't used, but is implemented like...

setFoo(a)
{
  if getFoo() == a
      return;
}

would only report setFoo as unused, so a check for "is both setFoo and
getFoo" unused and only report if it is, would leave both setFoo and
getFoo in there, even though both are actually really unused.

Already have a similar problem with

doFoo(a)
{
... doFoo(b)
}

i.e. something which is only used by itself is currently not reported
on, so cycles of uselessness are currently unreported.

C.