Completion stopped working

Completion stopped working

[Thorsten Jolitz]

Hi List, 

recently auto-completion stopped working completely for me, and I'm
trying to chase the root of this problem. 

Therefore I have 2 questions (at the moment):

1. C-h f and C-h v

With 'emacs -Q', these commands with point on a function or variable
(still) propose the symbol at point as a default, e.g.

,------------------------------------
| Describe function (default defvar):
`------------------------------------

but this stopped working when I load my .emacs. "Who" is in charge of
showing this default value?

2. auto-complete popup window

I remember in the back of my head that I disabled a (generic) option to
popup help windows or so - but I don't remember the name of this option
anymore and thus don't find it in my .emacs.

Does such an option exist and whats its name?
Could this be the reason that auto-complete-mode stopped working?

PS 1
System-info:

,----------------------------------------------------------------
| GNU Emacs 24.3.1 (x86_64-unknown-linux-gnu, GTK+ Version 3.6.4)
|  of 2013-03-18 on eric
| Org 8.0-pre
| Ma Gnus v0.2
`----------------------------------------------------------------

PS 2
my (simple) auto-complete-mode config which used to work:

,---------------------------------------------------------------------------
| ;; **** 26.8.4 Auto Complete Mode
| 
| ;; auto-complete-mode
| (when (require 'auto-complete nil 'NOERROR)
| 
| (add-to-list 'ac-dictionary-directories "~/.emacs.d/dict")
| (require 'auto-complete-config)
| (ac-config-default)
| 
| ;; more add-hook's in other places, e.g. for emacs lisp
| (add-hook 'LaTex-mode-hook lambda () (auto-complete-mode +1)))
| (add-hook 'org-mode-hook (lambda () (auto-complete-mode +1)))
| 
| ;; (define-key ac-menu-map "\C-n" 'ac-next)
| ;; (define-key ac-menu-map "\C-p" 'ac-previous)
| 
| (setq ac-auto-start 3)
| 
| (define-key ac-complete-mode-map "\M-/" 'ac-stop) )
| 
| ;; (define-key ac-complete-mode-map [f12] 'ac-complete)
| ;; (define-key ac-complete-mode-map "\r" nil))
`---------------------------------------------------------------------------

-- 
cheers,
Thorsten

Re: Completion stopped working

[Christopher Schmidt]

Thorsten Jolitz <tjolitz@gmail.com> writes:
> "Who" is in charge of showing this default value?

    function-called-at-point
    variable-at-point

If I were you, I'd try to bisect my init.el.  Use emacs -q and eval the
forms in your init.el one after another until you hit on your bug.

        Christopher

Re: Completion stopped working

[Thorsten Jolitz]

Christopher Schmidt <christopher@ch.ristopher.com> writes:

> Thorsten Jolitz <tjolitz@gmail.com> writes:
>> "Who" is in charge of showing this default value?
>
>     function-called-at-point
>     variable-at-point

I see, thanks

> If I were you, I'd try to bisect my init.el.  Use emacs -q and eval the
> forms in your init.el one after another until you hit on your bug.

the disadvantage of having a really big init.el is that this seems like
a lot of work, but anyway, I'll try ...

-- 
cheers,
Thorsten

Re: Completion stopped working

[Christopher Schmidt]

Thorsten Jolitz <tjolitz@gmail.com> writes:
> the disadvantage of having a really big init.el is that this seems
> like a lot of work, but anyway, I'll try ...

Perform a binary search with eval-region then.  ;)

        Christopher

Re: Completion stopped working

[Stefan Monnier]

> the disadvantage of having a really big init.el is that this seems like
> a lot of work, but anyway, I'll try ...

With "bisect" (aka binary search), even a very large file can be
analysed pretty quickly.  10 steps reduce the size of the search space
by 1000, 20 steps by a million.


        Stefan

Re: Completion stopped working

[Thorsten Jolitz]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> the disadvantage of having a really big init.el is that this seems like
>> a lot of work, but anyway, I'll try ...
>
> With "bisect" (aka binary search), even a very large file can be
> analysed pretty quickly.  10 steps reduce the size of the search space
> by 1000, 20 steps by a million.

Thats encouraging ;)
I'll give it a try. 

-- 
cheers,
Thorsten

Bug in help-fns+.el (was Re: Completion stopped working)

[Thorsten Jolitz]

Christopher Schmidt <christopher@ch.ristopher.com> writes:

> Thorsten Jolitz <tjolitz@gmail.com> writes:
>> "Who" is in charge of showing this default value?
>
>     function-called-at-point
>     variable-at-point
>
> If I were you, I'd try to bisect my init.el.  Use emacs -q and eval the
> forms in your init.el one after another until you hit on your bug.

Following your advice, I found out that loading 'help-fns+.el' by Drew
Adams redefines 'describe-function' and 'describe-variable', and fails
at offering the symbol near point as the default symbol to be described.

-- 
cheers,
Thorsten

RE: Bug in help-fns+.el (was Re: Completion stopped working)

[Drew Adams]

> Following your advice, I found out that loading 'help-fns+.el' by Drew
> Adams redefines 'describe-function' and 'describe-variable', and fails
> at offering the symbol near point as the default symbol to be 
> described.

Please report the problem you think you see to me, using a reproducible recipe
from emacs -Q.

Re: Bug in help-fns+.el (was Re: Completion stopped working)

[Thorsten Jolitz]

"Drew Adams" <drew.adams@oracle.com> writes:

>> Following your advice, I found out that loading 'help-fns+.el' by Drew
>> Adams redefines 'describe-function' and 'describe-variable', and fails
>> at offering the symbol near point as the default symbol to be 
>> described.
>
> Please report the problem you think you see to me, using a
> reproducible recipe from emacs -Q.

1. emacs -Q
2. type 

,------------------
| (message "hello")
`------------------

in scratch buffer

3. with point on 'message', do 'C-h f' and you'll see:

,-------------------------------------
| Describe function (default message):
`-------------------------------------

4. Do 'M-x load-file RET /path/to/help-fns+.el'

5. with point on 'message', do 'C-h f' again and you'll see:

,--------------------
| Describe function: 
`--------------------


PS:
GNU Emacs 24.3.1 (x86_64-unknown-linux-gnu, GTK+ Version 3.6.4)
 of 2013-03-18 on eric

-- 
cheers,
Thorsten

RE: Bug in help-fns+.el (was Re: Completion stopped working)

[Drew Adams]

> 1. emacs -Q
> 2. type 
>    (message "hello")
>    in scratch buffer
> 3. with point on 'message', do 'C-h f' and you'll see:
>    Describe function (default message):
> 4. Do 'M-x load-file RET /path/to/help-fns+.el'
> 5. with point on 'message', do 'C-h f' again and you'll see:
>    Describe function: 

Thanks for the quick reply.

What happens if you hit `M-n' - what default value is inserted in the
minibuffer?  (I see `message' inserted.)

What happens if you hit `RET' - what function is described by default?  (For me
it is `message'.)

IOW, from what I see, the only difference is whether the default value is shown
in parens.  Is that what you see also?  Is that the problem, as you see it, or
is there something else?  Thx.

Re: Bug in help-fns+.el (was Re: Completion stopped working)

[Thorsten Jolitz]

"Drew Adams" <drew.adams@oracle.com> writes:

>> 1. emacs -Q
>> 2. type 
>>    (message "hello")
>>    in scratch buffer
>> 3. with point on 'message', do 'C-h f' and you'll see:
>>    Describe function (default message):
>> 4. Do 'M-x load-file RET /path/to/help-fns+.el'
>> 5. with point on 'message', do 'C-h f' again and you'll see:
>>    Describe function: 

> What happens if you hit `M-n' - what default value is inserted in the
> minibuffer?  (I see `message' inserted.)

I see `message' too. 

> What happens if you hit `RET' - what function is described by default?
> (For me it is `message'.)

For me  it is `message' too. 

> IOW, from what I see, the only difference is whether the default value
> is shown in parens. Is that what you see also? Is that the problem, as
> you see it, or is there something else? Thx.

Yes, it seems this is the only problem - more an inconvenience than a
real bug. It would be nicer to see the default value in parens, but its
not a big issue then. 

-- 
cheers,
Thorsten

RE: Bug in help-fns+.el (was Re: Completion stopped working)

[Drew Adams]

> > IOW, from what I see, the only difference is whether the 
> > default value is shown in parens.  Is that what you see also?
> > Is that the problem, as you see it, or is there something
> > else?  Thx.
> 
> Yes, it seems this is the only problem - more an inconvenience
> than a real bug. It would be nicer to see the default value in 
> parens, but its not a big issue then. 

FYI - I went ahead and added the default back to the prompt.

(If you use Icicles then you can choose how default values are handled - wrt the
prompt and wrt minibuffer insertion.  IOW, you decide, not the individual
command.)

Re: Bug in help-fns+.el (was Re: Completion stopped working)

[Thorsten Jolitz]

"Drew Adams" <drew.adams@oracle.com> writes:

>> > IOW, from what I see, the only difference is whether the 
>> > default value is shown in parens.  Is that what you see also?
>> > Is that the problem, as you see it, or is there something
>> > else?  Thx.
>> 
>> Yes, it seems this is the only problem - more an inconvenience
>> than a real bug. It would be nicer to see the default value in 
>> parens, but its not a big issue then. 
>
> FYI - I went ahead and added the default back to the prompt.

Thanks, I just downloaded the new version and the prompt shows the
default again. 

> (If you use Icicles then you can choose how default values are handled
> - wrt the prompt and wrt minibuffer insertion. IOW, you decide, not
> the individual command.)

I actually use ido-mode, and had a look at 'helm' recently and found it
quite interesting. In that area, Emacs really offers functionality
overkill, its hard to decide what to use. 

-- 
cheers,
Thorsten