Re: Calling completion function from a non-completion ZLE widget

Re: Calling completion function from a non-completion ZLE widget

[Sven Wischnowsky]

Bart Schaefer wrote:

> ...
> 
> The best we can do seems to be a bit of defensive programming up front.
> What's a reliable way to determine that the completion system is active?
> (Testing a parameter won't do it, the user could always set another of
> the same name.)

Hm, no parameter? Even if ${(t)compstate} yields `association-local-special'?

We can use the conditions, e.g.:

  if [[ -prefix * ]] 2> /dev/null; then
    # completion active
  fi

Or should we change compadd to return `2' in $? when completion is not 
active? Or `1' there, but `2' if no matches are given, so that we
could do

  compadd 2> /dev/null
  if [[ $? -ne 1 ]]; then
    ...
  fi

? Dunno.

Bye
 Sven


--
Sven Wischnowsky                         wischnow@informatik.hu-berlin.de

Re: Calling completion function from a non-completion ZLE widget

[Bart Schaefer]

On May 30,  9:59am, Sven Wischnowsky wrote:
}
} Bart Schaefer wrote:
} 
} > What's a reliable way to determine that the completion system is active?
} > (Testing a parameter won't do it, the user could always set another of
} > the same name.)
} 
} Hm, no parameter? Even if ${(t)compstate} yields `association-local-special'?

Ah, I forgot about (t).  Yes, that would probably suffice.

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net   

Calling completion function from a non-completion ZLE widget

[Bart Schaefer]

I accidentally ended up with a function bound to an ordinary widget (i.e.
zle -N ...) that called _complete_help. [*]  Bad, very bad.  Not only did
it complain:

_main_complete:compset:52: can only be called from completion function
_default:compcall:12: can only be called from completion function
_main_complete:266: bad set of key/value pairs for associative array

But now that shell is also in a state where no completion of any kind does
anything other than beep at me and occasionally spit out errors about
bad math expressions, like this:

zstyle:5: bad math expression: operand expected at `:completio...'

Hmm, I suspect _complete_help got aborted in progress by one of the above
errors and left some functions defined.  Yup, that's it; probably left
some other garbage state as well.  Yow, it even left $compstate in an
inconsistent state!  $+compstate is 1, $#compstate is also 1, but trying
`typeset | grep ^compstate` yields nothing.

It'd be best if that sort of thing doesn't happen, but there's no way to
trap on only the sort of error that causes the whole call stack to abort
(as `bad set of ...' seems to have).

The best we can do seems to be a bit of defensive programming up front.
What's a reliable way to determine that the completion system is active?
(Testing a parameter won't do it, the user could always set another of
the same name.)

[*] It should have been calling `zle _complete_help', I forgot the `zle'.

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net