PATCH: ZLS_COLORS/ZLS_COLOURS consistency in _setup

PATCH: ZLS_COLORS/ZLS_COLOURS consistency in _setup

[Clint Adams]

This attempts to follow the course of action outlined as option #1
in the message included after the patch, though I agree that clobbering
those parameters is counterintuitive.

I wonder why ZLS_COLORS and ZLS_COLOURS aren't linked together
as special parameters by zsh/complist.  I also think the precedence
of those two should be documented.

Index: Completion/Base/Core/_setup
===================================================================
RCS file: /cvsroot/zsh/zsh/Completion/Base/Core/_setup,v
retrieving revision 1.1
diff -u -r1.1 _setup
--- Completion/Base/Core/_setup	2001/04/02 11:04:32	1.1
+++ Completion/Base/Core/_setup	2001/05/28 15:42:33
@@ -20,7 +20,7 @@
 #   ZLS_COLORS="$ZLS_COLORS$ZLS_COLOURS"
 
 elif [[ "$1" = default ]]; then
-  unset ZLS_COLORS
+  unset ZLS_COLORS ZLS_COLOURS
 fi
 
 if zstyle -t ":completion:${curcontext}:$1" list-packed; then
Index: Doc/Zsh/compsys.yo
===================================================================
RCS file: /cvsroot/zsh/zsh/Doc/Zsh/compsys.yo,v
retrieving revision 1.124
diff -u -r1.124 compsys.yo
--- Doc/Zsh/compsys.yo	2001/05/09 08:55:59	1.124
+++ Doc/Zsh/compsys.yo	2001/05/28 15:42:36
@@ -1483,7 +1483,7 @@
 item(tt(list-colors))(
 If the tt(zsh/complist) module is used, this style can be used to set
 color specifications as with the tt(ZLS_COLORS) and tt(ZLS_COLOURS)
-parameters (see
+parameters, which will not be honored under this completion system (see
 ifzman(the section `The zsh/complist Module' in zmanref(zshmodules))\
 ifnzman(noderef(The zsh/complist Module))\
 ).
Index: Doc/Zsh/mod_complist.yo
===================================================================
RCS file: /cvsroot/zsh/zsh/Doc/Zsh/mod_complist.yo,v
retrieving revision 1.14
diff -u -r1.14 mod_complist.yo
--- Doc/Zsh/mod_complist.yo	2000/10/05 08:41:37	1.14
+++ Doc/Zsh/mod_complist.yo	2001/05/28 15:42:36
@@ -127,7 +127,7 @@
 `tt($colors[red])' to get the code for foreground color red and
 `tt($colors[bg-green])' for the code for background color green.
 
-If the completion system based around shell functions is used, these
+If the completion system invoked by compinit is used, these
 parameters should not be set directly because the system controls them 
 itself.  Instead, the tt(list-colors) style should be used (see
 ifzman(the section `Completion System Configuration' in zmanref(zshcompsys))\

----- Forwarded message from Matt Zimmerman <mdz@debian.org> -----

> In that case, do you get the same behavior with zsh -f ?
> What's causing the zsh/complist module to be loaded?
> Is it direct, or is it due to a zstyle setting?

The problem seems to have to do with when compinit is loaded.  For example,
when I do this:

mizar:[~] env -i zsh -f
mizar% export TERM=xterm
mizar% eval `dircolors -b`
mizar% export ZLS_COLORS="$LS_COLORS"
mizar% zmodload zsh/complist
mizar% cd [TAB]

I get a colored completion listing.  However, if I load compinit (mimicking
what happens in my .zlogin):

mizar:[~] env -i zsh -f
mizar% export TERM=xterm
mizar% eval `dircolors -b`
mizar% export ZLS_COLORS="$LS_COLORS"
mizar% zmodload zsh/complist
mizar% autoload -U compinit
mizar% compinit
mizar% cd [TAB]

the listing is not colored.  However, if I use the ZLS_COLOURS variable
instead:

mizar:[~] env -i zsh -f
mizar% export TERM=xterm
mizar% eval `dircolors -b`
mizar% export ZLS_COLOURS="$LS_COLORS"
mizar% zmodload zsh/complist
mizar% autoload -U compinit;compinit
mizar% cd [TAB]

the listing is colored.  This is what prompted me to file this bug report
originally.  Even if I set the variable after compinit is loaded:

mizar:[~] env -i zsh -f
mizar% export TERM=xterm
mizar% eval `dircolors -b`
mizar% zmodload zsh/complist
mizar% autoload -U compinit;compinit
mizar% export ZLS_COLORS="$LS_COLORS"

I get no colors.  However, I now notice this comment at the bottom of the
complist section is zshmodules(1):

       If  the  completion system based around shell functions is
       used, these parameters should not be set directly  because
       the system controls them itself.  Instead, the list-colors
       style should be used (see the section  `Completion  System
       Configuration' in zshcompsys(1)).

I didn't notice this section because it refers to "the completion system based
around shell functions", which didn't make me think of the new completion
system.  Maybe it should be clarified.  So, it looks like neither _COLORS or
_COLOURS should work when using compinit, but _COLOURS still does.  That is a
software bug.

There is another documentation bug: what prompted me to set these variables
originally was this passage in zshcompsys(1):

       list-colors
              If  the zsh/complist module is used, this style can
              be used to set color  specifications  as  with  the
              ZLS_COLORS and ZLS_COLOURS parameters (see the sec­
              tion `The zsh/complist Module' in zshmodules(1)).

This section should make it clear that ZLS_* should _not_ be used, and will be
ignored.  As it is, it implies that they will still be honored.

In summary:

- Either:
  1. Both ZLS_COLOURS and ZLS_COLORS should be ignored when compinit is loaded,
    not just ZLS_COLORS
  2. Both variables should be used (I find this behavior more intuitive)

- If 1., zshcompsys(1) should mention that these variables will be ignored

- If 1., zshmodules(1) should be clarified to say which completion system
  it is talking about (perhaps using wording like "the completion system
  described in zshcompsys(1)", if there is no proper name for it)

----- End forwarded message -----

Re: PATCH: ZLS_COLORS/ZLS_COLOURS consistency in _setup

[Sven Wischnowsky]

Clint Adams wrote:

> This attempts to follow the course of action outlined as option #1
> in the message included after the patch, though I agree that clobbering
> those parameters is counterintuitive.

Hm. I'm almost constantly amazed by all the things we people consider
`intuitive' these days ;-)

But anyway... there is no way around this if we want to be able to
define context-specific colors. Which we want.

> I wonder why ZLS_COLORS and ZLS_COLOURS aren't linked together
> as special parameters by zsh/complist. 

Probably mostly simplicity in the C-code.  These two aren't special in
any way.  The code just looks them up.

> I also think the precedence
> of those two should be documented.

Hm, they are mentioned in the order in which they are looked up[1].


Bye
  Sven

[1] So `COLORS' comes first.  And that from me.  I would have guessed
    it were the other way round.  How counterintuitive.

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