[ESS] drop empty help buffers

[Martin Maechler]

>>>>> Rodney Sparapani <rsparapa at mcw.edu>
>>>>>     on Tue, 28 Feb 2012 16:12:01 -0600 writes:

    > Martin Maechler wrote:
    >> that was fast.
    >> 
    >> IIRC the previous default was quite on purpose:
    >> 
    >> It's not only for a non-existing command that you get a
    >> bogous help buffer, and in some of other such cases, the
    >> user should be able to see and read what the bogous
    >> buffer contains.  (and also note that error messages by
    >> in other languages than English, so you can't just grep
    >> for the kind of error message,...)
    >> 
    >> Martin
    >> 
    > Well, the buffer contains...  No documentation for 'foo'
    > in specified packages and libraries: you could try '??foo'

    > And ??foo gives...  No documentation for '?foo' in
    > specified packages and libraries: you could try '???foo'
only in ESS.  That's the infamous "?"-ESS-bug "box"
we had encountered many times.
in an R outside of ESS "??"  'works'

    > What can you learn from that?  

    > With t as the default; it
    > just says "No documentation" in the mini-buffer.  What am
    > I missing?

the "??" (and the  <foo>?<bar>  and  ?call(..)  
smart help invocations that ESS still does not support..
{see ESS-core e-mails in the past}.

*AND* (I think): I said "in some of other such cases",
so I'm pretty sure we had other cases of bogous help buffers
(e.g., when you where in debug(.) mode, or when I as R developer
 had debug(help) active, and ...)
So there are/were rare cases, where a user remained really
clueless if she could not see the bogous help buffer.

Martin