[logon] (somewhat fundamental) updates in the LOGON tree

[Berthold Crysmann]

On Fri, 2009-06-05 at 20:22 +0200, Stephan Oepen wrote:
> hi berthold (and dan),
> 
> > I discovered a problem with generation from predicate types (not
> > strings).  While the lkb.old tree happily generates from such
> > predicates, the new branch complains about unknown predicates.

> my apologies, i had introduced a bug in lexial instantiation.  the bug
> was triggered by predicates that are used both in lexical entries and
> grammar rules (i.e. only a sub-set of non-string predicates), thus was
> not exposed in the limited amount of testing i had done yesterday.
> 

That's congruent with my observations: I only observe this with
dem_q_rel which appears bothy in lexical rules (bound nan/can) and in
free forms (wannan wancan etc.). No problem generating the bound form
and no problem with _some_q_rel which only has free form realisations.

> dan, could you confirm this works for you, and then i re-build binaries
> for berthold tomorrow?
> 
>                                                        cheers  -  oe
> 
> nb: ann, this is exactly about the code we discussed yesterday.  i have
> now resorted to the latest variant i had proposed in that discussion:
> 
>   (defun lexical-rel-p (pred)
>     (or (gethash pred *relation-index*)
>         (and (stringp pred) (> (length pred) 0) 
>              (char= (char pred 0) #\_))))
> 
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> +++ Universitetet i Oslo (IFI); Boks 1080 Blindern; 0316 Oslo; (+47) 2284 0125
> +++     CSLI Stanford; Ventura Hall; Stanford, CA 94305; (+1 650) 723 0515
> +++       --- oe at ifi.uio.no; oe at csli.stanford.edu; stephan at oepen.net ---
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++