Re: Getting completion to tell the user what to do

[Sven Wischnowsky <wischnow@xxxxxxxxxxxxxxxxxxxxxxx>]

Bart Schaefer wrote:

> On Jul 20,  9:08am, Sven Wischnowsky wrote:
> } Subject: Re: Getting completion to tell the user what to do
> }
> } Bart Schaefer wrote:
> } 
> } > What I want is to offer no completions at all [...]
> } > but print a hint to the user as to what he's supposed to type.
> } > 
> } > What obvious thing have I forgotten/overlooked here?  Is there an entirely
> } > better alternative to using compadd -X ?
> } 
> } Time for dirty tricks:
> } 
> }   compadd -UX 'Please...' -n ''
> }   compstate[insert]=''
> 
> Oho -- undocumented dirty tricks, no less.  The compstate[insert] doc
> doesn't give any hint that it can be set but empty.  Of which case that
> is documented is this a degenerate?

Unsetting compstate[insert].

> }   compstate[list]=list
> }   compstate[force_list]=yes
> } 
> } You need the -U because otherwise the empty string never matches
> } what's on the line (not even the empty string on the line).
> 
> Really?  When I first tried
> 
> 	compadd -X 'Please ...' ''
> 
> (without the -S) then every time I pressed TAB a single space got inserted
> (the suffix).  If the empty string isn't a match, why did that happen?

Ugh, right, sorry.

> } Then we can switch off insertion completely.
> 
> So that's what compstate[insert]='' means?

Exactly that.

> } If you want the string to be listed only on a TAB with an empty string 
> } you can do `compadd -X "Please..." -n dummy' -- i.e. add a string with 
> } matching.
> 
> Presumably then "dummy" should be something that can't possibly match?
> Or does that not matter?

That would be better, yes. Unless you also have `compstate[insert]=""' or
`unset "compstate[insert]"', in which case it doesn't matter.

> } The -n, of course, is just to be sure that the matches are not visible.
> } 
> } Is that good enough?
> 
> It's just marvelous, thanks.  One question, though -- I thought that we
> made kill-whole-line erase the completion listing, but it doesn't seem to
> do that in this case:
> 
>     zagzig% mail -s <TAB><C-u>
> 
> leaves me with
> 
>     zagzig% 
>     Please enter a descriptive subject
> 
> which is rather annoying when I start in typing some completely different
> command and doing completions for it that don't themselves produce a list.

Nasty side-effect of 7161.

Bye
 Sven

diff -u os/Zle/zle_tricky.c Src/Zle/zle_tricky.c
--- os/Zle/zle_tricky.c	Tue Jul 20 10:42:27 1999
+++ Src/Zle/zle_tricky.c	Tue Jul 20 11:08:31 1999
@@ -6743,8 +6743,9 @@
 	listmatches();
     if (validlist)
 	freematches();
-    lastambig = menucmp = menuacc = validlist = showinglist =
-	fromcomp = listshown = 0;
+    lastambig = menucmp = menuacc = validlist = showinglist = fromcomp = 0;
+    if (listshown < 0)
+	listshown = 0;
     minfo.cur = NULL;
     minfo.asked = 0;
     compwidget = NULL;
diff -u od/Zsh/compwid.yo Doc/Zsh/compwid.yo
--- od/Zsh/compwid.yo	Mon Jul 19 15:54:31 1999
+++ Doc/Zsh/compwid.yo	Tue Jul 20 11:14:45 1999
@@ -236,7 +236,8 @@
 behaviour of the tt(MENU_COMPLETE) or tt(AUTO_MENU) options, respectively,
 is to be used.
 
-On exit it may be set to any of the values above, or to a number, in which
+On exit it may be set to any of the values above (where setting it to
+the empty string is the same as unsetting it), or to a number, in which
 case the match whose number is given will be inserted into the command line.
 It may also be set to a string of the form `var(group):var(match)' which
 specifies a match from a group of matches to be inserted, counting from 1

--
Sven Wischnowsky                         wischnow@xxxxxxxxxxxxxxxxxxxxxxx