Re: Preventing sorting in completers

[Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>]

On Jul 14,  6:31pm, Peter Stephenson wrote:
} Subject: Re: Preventing sorting in completers
}
} Bart Schaefer wrote:
} > cases where the internal sort done by the completion system does not
} > agree with the sort performed by ${(o)exp}.
} 
} I assumed that (o) was equivalent to what the completion system did.

It may very well be, and most likely is.
 
} > _expand uses
} >     zstyle -s ":completion:${curcontext}:" sort sort
} > whereas with the patch _description uses
} >     zstyle -s ":completion:${curcontext}:$1" sort sort
} 
} Well, the only ways I can see of completely removing problems like this
} are to pass down an additional argument into _description and _wanted to
} stop it using the style, or use a different style.

These aren't necessarily problems to be removed, just behaviors to be
documented.  There's no reason one can't have a style for sorting the
results of _expand and another more specific style for sorting a tag
context.

OTOH, I wonder whether _description should NOT use the style when an
explicit -J or -V argument has been passed?  Most of the time there is
no -J/-V argument at all to _description.