RE: predictive.el -- predictive completion of words as you type inEmacs

[Phillip Lord]

Thorsten Bonow wrote:
>>>>>> "Uwe" == Uwe Brauer <address@hidden> writes:
> 
>     Uwe> There is also dabbrev-hover.el, however I had difficulties
>     with this Uwe> package under Xemacs, however your pkg seem to
>     work nicely. Predictive Uwe> on the other hand seems also not to
>     work to smoothly with Xemacs (so Uwe> far).
> 
> (Off topic I guess, maybe we should move to an XEmacs group...)
> 
> Hm, I can't get pabbrev.el to work with XEmacs. I'm using XEmacs
> 21.4.19 coming 
> with my up to date Debian Unstable. Starting with xemacs -vanilla and
> just 
> loading pabbrev.el (after setting debug-on-error to t) gives me an
> error on the 
> first completion.
> 
> Did you do something special to make it work? Using 21.5 or something?
> 

> The *Backtrace* buffer gives me:
> 
> Debugger entered--Lisp error: (error "Error in pabbrev-mode")
>   signal(error ("Error in pabbrev-mode"))
>   cerror("Error in pabbrev-mode")
>   apply(cerror "Error in pabbrev-mode" nil)
>   error("Error in pabbrev-mode")
>   pabbrev-command-hook-fail((void-function make-overlay) "post")



make-overlay is in the fsf-compatibility package, I think. I don't have
an xemacs around to try, and find it hard to use as I'm more used to 
emacs. I try to get these differences sorted out, but it's not easy!
There are so many little differences. I wasn't surprised to find
differences
in make-overlay as its "emulated" on xemacs. But did you know that
the `subseq' function behaves differently between the two!

Phil