question about "-E" option of compadd

question about "-E" option of compadd

[Han Pingtian]

Hi there,

There is one statement about -E option of compadd in the manual:

    And because all but one empty string would otherwise be removed, 
    this  option implies  the  -V  and  -2 options (even if an 
    explicit -J option is given).

But isn't the action "all but one empty string be removed" implying a
"-1" option? "-2" option should keep all empty strings I think.

Thanks in advance!

Re: question about "-E" option of compadd

[Bart Schaefer]

On Sun, Apr 24, 2016 at 8:07 PM, Han Pingtian <hanpt@linux.vnet.ibm.com> wrote:
>
>     And because all but one empty string would otherwise be removed,
>     this  option implies  the  -V  and  -2 options (even if an
>     explicit -J option is given).
>
> But isn't the action "all but one empty string be removed" implying a
> "-1" option? "-2" option should keep all empty strings I think.

Note the word "otherwise" in the "And because ..." sentence.
"Otherwise" in this context means "if -E were not used".

"And because all but one empty string would be removed if -E were not
used, it is the case that -E implies -V and --2."

Re: question about "-E" option of compadd

[Han Pingtian]

On Sun, Apr 24, 2016 at 09:36:00PM -0700, Bart Schaefer wrote:
> On Sun, Apr 24, 2016 at 8:07 PM, Han Pingtian <hanpt@linux.vnet.ibm.com> wrote:
> >
> >     And because all but one empty string would otherwise be removed,
> >     this  option implies  the  -V  and  -2 options (even if an
> >     explicit -J option is given).
> >
> > But isn't the action "all but one empty string be removed" implying a
> > "-1" option? "-2" option should keep all empty strings I think.
> 
> Note the word "otherwise" in the "And because ..." sentence.
> "Otherwise" in this context means "if -E were not used".
> 
> "And because all but one empty string would be removed if -E were not
> used, it is the case that -E implies -V and --2."

Thanks. But looks like even if -E was used, all but one empty strings
still being removed ? like this:

% _foo()
compadd -E 3 foo1 foo1 foo1 foo2 foo3 '' '' ''
% compdef _foo foo
% foo <tab>
   foo1 foo2 foo3


there is only one empty string showed.

Thanks in advance.

Re: question about "-E" option of compadd

[Bart Schaefer]

On Apr 25,  3:48pm, Han Pingtian wrote:
} Subject: Re: question about "-E" option of compadd
}
} But looks like even if -E was used, all but one empty strings
} still being removed ? like this:
} 
} % _foo()
} compadd -E 3 foo1 foo1 foo1 foo2 foo3 '' '' ''
} % compdef _foo foo
} % foo <tab>
}    foo1 foo2 foo3

There's this other part of the documentation for -E

    empty matches [are] only useful to format completion lists and to
    make explanatory string be shown in completion lists 

Without -V -2, the completion listing is deduplicated on the match
string regardless of the descriptions, so you'll only see one of the
possibly several descriptions.  With -V -2, identical matches and
their corresponding descriptions are shown, but if the descriptions
are missing then the deduplication still happens.

In other words -E is only used to "pad" the list of matches to have
the same number of elements as the array of descriptions passed to -d.
It's not an error to pass -E without -d but it's not useful either.

Or at least that's my understanding of how it's expected to work, and
is how it works in e.g. Oliver's new _dates helper function; but it
is quite possible that some other combination of options bends it.

Re: question about "-E" option of compadd

[Han Pingtian]

On Mon, Apr 25, 2016 at 12:20:04PM -0700, Bart Schaefer wrote:
> On Apr 25,  3:48pm, Han Pingtian wrote:
> } Subject: Re: question about "-E" option of compadd
> }
> } But looks like even if -E was used, all but one empty strings
> } still being removed ? like this:
> } 
> } % _foo()
> } compadd -E 3 foo1 foo1 foo1 foo2 foo3 '' '' ''
> } % compdef _foo foo
> } % foo <tab>
> }    foo1 foo2 foo3
> 
> There's this other part of the documentation for -E
> 
>     empty matches [are] only useful to format completion lists and to
>     make explanatory string be shown in completion lists 
> 
> Without -V -2, the completion listing is deduplicated on the match
> string regardless of the descriptions, so you'll only see one of the
> possibly several descriptions.  With -V -2, identical matches and
> their corresponding descriptions are shown, but if the descriptions
> are missing then the deduplication still happens.
> 
> In other words -E is only used to "pad" the list of matches to have
> the same number of elements as the array of descriptions passed to -d.
> It's not an error to pass -E without -d but it's not useful either.
> 
Thanks so much, now I got it!
> Or at least that's my understanding of how it's expected to work, and
> is how it works in e.g. Oliver's new _dates helper function; but it
> is quite possible that some other combination of options bends it.