Re: bindkey bug with zsh-3.1.6-dev-21?

Re: bindkey bug with zsh-3.1.6-dev-21?

[Sven Wischnowsky]

Peter Stephenson wrote:

> > Ick. The problem is that in autoload_zleread() we do:
> > 
> >     if (load_module("zsh/zle"))
> > 	load_module("zsh/compctl");
> > 
> > i.e., we attempt to load compctl only if loading zle succeeds. For
> > people who have a call to `bindkey' in their .z*, the zle module will
> > already be loaded, re-loading fails and compctl is not loaded. Bummer.
> > 
> > Can anyone think of a solution for this? I can't (apart from trying to 
> > load compctl unconditionally, which I don't like).
> 
> Couldn't we also check whether ((trashzleptr == noop_function))?  If it is,
> zle isn't loaded, and if it isn't, zle must be loaded or something very
> weird has happened.  zlereadptr is even declared in both branches of the
> #ifdef just above.

Right, but what I would like to avoid is to load compctl whenever zle
is loaded. Otherwise anyone using the new completion system (which
loads zle indirectly) will have ol' compctl hanging around. And
_default will then try compctl's etc.

Hm, maybe a test of the form

  if (load(zle) || !loaded(complete)) load(compctl);

There's find_module() for that kind of test.

Does that sound sensible?

Bye
 Sven


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

Re: bindkey bug with zsh-3.1.6-dev-21?

[Peter Stephenson]

> Hm, maybe a test of the form
> 
>   if (load(zle) || !loaded(complete)) load(compctl);
> 
> There's find_module() for that kind of test.
> 
> Does that sound sensible?

Looks worth trying --- assuming (which I haven't checked) that something in
compinit loads complete, this is about the best you can do.  The only other
possibilitity I can see is to delay loading old completion until zle
actually starts up --- then if there's no new completion, it's safe to
assume old completion is required.  That might be quite sensible if it's
easy to implement.

-- 
Peter Stephenson <pws@cambridgesiliconradio.com>
Cambridge Silicon Radio, Unit 300, Science Park, Milton Road,
Cambridge, CB4 0XL, UK                          Tel: +44 (0)1223 392070

Re: bindkey bug with zsh-3.1.6-dev-21?

[Sven Wischnowsky]

Peter Stephenson wrote:

> > Hm, maybe a test of the form
> > 
> >   if (load(zle) || !loaded(complete)) load(compctl);
> > 
> > There's find_module() for that kind of test.
> > 
> > Does that sound sensible?
> 
> Looks worth trying --- assuming (which I haven't checked) that something in
> compinit loads complete, this is about the best you can do. 

`zle -C' load zsh/complete.

> The only other
> possibilitity I can see is to delay loading old completion until zle
> actually starts up --- then if there's no new completion, it's safe to
> assume old completion is required.  That might be quite sensible if it's
> easy to implement.

Sounds good too...

I've no time to hack now...

Bye
 Sven


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

Re: bindkey bug with zsh-3.1.6-dev-21?

[Peter Stephenson]

> Ick. The problem is that in autoload_zleread() we do:
> 
>     if (load_module("zsh/zle"))
> 	load_module("zsh/compctl");
> 
> i.e., we attempt to load compctl only if loading zle succeeds. For
> people who have a call to `bindkey' in their .z*, the zle module will
> already be loaded, re-loading fails and compctl is not loaded. Bummer.
> 
> Can anyone think of a solution for this? I can't (apart from trying to 
> load compctl unconditionally, which I don't like).

Couldn't we also check whether ((trashzleptr == noop_function))?  If it is,
zle isn't loaded, and if it isn't, zle must be loaded or something very
weird has happened.  zlereadptr is even declared in both branches of the
#ifdef just above.

-- 
Peter Stephenson <pws@cambridgesiliconradio.com>
Cambridge Silicon Radio, Unit 300, Science Park, Milton Road,
Cambridge, CB4 0XL, UK                          Tel: +44 (0)1223 392070

Re: bindkey bug with zsh-3.1.6-dev-21?

[Thomas Köhler]

[-- Attachment #1: Type: text/plain, Size: 2263 bytes --]

On Wed, Apr 19, 2000 at 10:58:14AM +0200,
Sven Wischnowsky <wischnow@informatik.hu-berlin.de> wrote:
> 
> =?iso-8859-1?Q?Thomas_K=F6hler?= wrote:
> 
> > Hi,
> > Seems that there's a problem with bindkey and zsh-3.1.6-dev-21 and
> > zsh-3.1.6-dev-22:
> > 
> > When I use this .zshrc:
> > 
> >   bindkey -e
> >   export EDITOR=3Dvi
> > 
> > I get not tab-completion. If I type "bindkey", I get a long list of
> > bindkeys, amongst others this one:
> > "^I" expand-or-complete
> > 
> > So, the binding is there, but it just doesn't work :-(
> > 
> > The funny thing is, when I just have the "export EDITOR=3Dvi" line in my
> > =2Ezshrc and type "bindkey -e" on the prompt later, tab-completion works.
> > How to resolve the problem?
> 
> Put `zmodload zsh/compctl' in your `.zshrc'. Or better, use the new
> completion system (autoloading and calling `compinit').

Well, *I* do use the new completion system, but the one who found the
problem (tries zsh because I showed him what it can do) doesn't (yet?).

> Ick. The problem is that in autoload_zleread() we do:
> 
>     if (load_module("zsh/zle"))
> 	load_module("zsh/compctl");
> 
> i.e., we attempt to load compctl only if loading zle succeeds. For
> people who have a call to `bindkey' in their .z*, the zle module will
> already be loaded, re-loading fails and compctl is not loaded. Bummer.

Uhm. Bad.

> Can anyone think of a solution for this? I can't (apart from trying to 
> load compctl unconditionally, which I don't like).

Is there a test whether or not a module is loaded?
Something like
   
    if (load_module("zsh/zle") || is_loaded_module("zsh/zle"))
       load_module("zsh/compctl");

If there is no such function (I don't have any insight in zsh's source)
perhaps now is the time to implement one :-)

Other possible solution: load zsh/compctl if there's a bindkey in .z* -
but that's of course not very intuitive.

> Bye
>  Sven

CU,
Thomas

-- 
 Thomas Köhler Email:   jean-luc@picard.franken.de     | LCARS - Linux
     <><        WWW:     http://jeanluc-picard.de      | for Computers
                IRC:             jeanluc               | on All Real
               PGP public key available from Homepage! | Starships

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: bindkey bug with zsh-3.1.6-dev-21?

[Sven Wischnowsky]

=?iso-8859-1?Q?Thomas_K=F6hler?= wrote:

> Hi,
> Seems that there's a problem with bindkey and zsh-3.1.6-dev-21 and
> zsh-3.1.6-dev-22:
> 
> When I use this .zshrc:
> 
>   bindkey -e
>   export EDITOR=3Dvi
> 
> I get not tab-completion. If I type "bindkey", I get a long list of
> bindkeys, amongst others this one:
> "^I" expand-or-complete
> 
> So, the binding is there, but it just doesn't work :-(
> 
> The funny thing is, when I just have the "export EDITOR=3Dvi" line in my
> =2Ezshrc and type "bindkey -e" on the prompt later, tab-completion works.
> How to resolve the problem?

Put `zmodload zsh/compctl' in your `.zshrc'. Or better, use the new
completion system (autoloading and calling `compinit').


Ick. The problem is that in autoload_zleread() we do:

    if (load_module("zsh/zle"))
	load_module("zsh/compctl");

i.e., we attempt to load compctl only if loading zle succeeds. For
people who have a call to `bindkey' in their .z*, the zle module will
already be loaded, re-loading fails and compctl is not loaded. Bummer.

Can anyone think of a solution for this? I can't (apart from trying to 
load compctl unconditionally, which I don't like).

Bye
 Sven


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

bindkey bug with zsh-3.1.6-dev-21?

[Thomas Köhler]

[-- Attachment #1: Type: text/plain, Size: 825 bytes --]

Hi,
Seems that there's a problem with bindkey and zsh-3.1.6-dev-21 and
zsh-3.1.6-dev-22:

When I use this .zshrc:

  bindkey -e
  export EDITOR=vi

I get not tab-completion. If I type "bindkey", I get a long list of
bindkeys, amongst others this one:
"^I" expand-or-complete

So, the binding is there, but it just doesn't work :-(

The funny thing is, when I just have the "export EDITOR=vi" line in my
.zshrc and type "bindkey -e" on the prompt later, tab-completion works.
How to resolve the problem?

CU,
Thomas

-- 
 Thomas Köhler Email:   jean-luc@picard.franken.de     | LCARS - Linux
     <><        WWW:     http://jeanluc-picard.de      | for Computers
                IRC:             jeanluc               | on All Real
               PGP public key available from Homepage! | Starships

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]