Re: (debug) backtrace: how to show source code at current location.

[Larry Evans]

On 01/05/11 17:31, Tim X wrote:
> Larry Evans <cppljevans@suddenlink.net> writes:
> 
>> On 01/05/11 13:06, Drew Adams wrote:
>>>> Please, how can I find out what's causing this error message:
>>>>
>>>>   "(wrong-type-argument arrayp tab)"
>>>
>>> From your error message alone, it seems that the symbol `tab' is being 
>>> passed to
>>> some function that expects an array.  That's not much help yet, but things 
>>> might
>>> become clearer when you dig deeper using the debugger.
>>>
>>> (setq debug-on-error t)
>>>
>>> 1. Load source (*.el) files for the library where the error occurred.
>>>
>>> IOW, look at the `debug-on-error' backtrace to see which function(s) were
>>> involved when the error occurred.  Then use `C-h f' or `find-func' to find 
>>> the
>>> function definition(s) (source file(s)).  Load the source file(s) (*.el) 
>>> using
>>> `load-file' or `load-library', including the extension `.el' in the name.
>>>
>>> 2. Provoke the error and backtrace again.  It should now show more 
>>> source-code
>>> entry points instead of byte-code (the `(byte-code "ÁÂÃ...' stuff)
>>>
>>> 3. You can also use `M-x debug-on-entry THE-FUNCTION' to enter the debugger
>>> whenever a problematic function is invoked.  Then you can walk through the
>>> invocation to see more about what caused the error.
>>>
>>>
>>>
>> Thank you very much for the help Drew.
>> What I did was started with just:
>>
>>   (setq debug-on-error t)
>>
>> as shown in the attachment (a revision of the one posted earlier).
>> The started emacs; however, no backtrace appeared.  OTOH, the
>> *Messages* buffer contained:
>>
>>> Loading /home/evansl/.emacs.d/init.common.el (source)...
>>> Loading /home/evansl/.recentf...done
>>> Cleaning up the recentf list...done (0 removed)
>>> Loading /usr/share/emacs/site-lisp/haskell-mode/haskell-site-file.el 
>>> (source)...done
>>> Loading /home/evansl/.emacs.d/init.common.el (source)...done
>>> Loading desktop...done
>>> Loading eldoc...done
>>> uncompressing indent.el.gz...done
>>> Note: file is write protected
>>> Loading cc-mode...done
>>> Note: file is write protected
>>> c-tab-stop-my-list message
>>> c++-mode-my-hook message:tabs=
>>> (0 2 4)
>>> c++-mode-my-hook message{
>>> File mode specification error: (wrong-type-argument arrayp tab)
>>> Loading /usr/share/emacs/site-lisp/haskell-mode/haskell-mode.el 
>>> (source)...done
>>
>> Is there something else I should do to get a backtrace?
>>
> 
> Yep, start emacs as 
> 
> emacs --debug-init
> 
> Tim
> 
Thanks, Tim and Drew.
I did gunzip all the *.el.gz files in:

  /usr/share/emacs/23.1/lisp

and did:

  emacs --debug-init

However, I never could get the emacs debugger to
show me the actual line it was on when I issued the debug
commands, such as d or c.

Anyway, one time the top line in the *Backtrace* buffer
showed a (define-key ...) expression whose tail end included
the args to a (define-key...) from init.common.el file:

     (define-key c++-mode-map 'tab 'c-insert-my-tab)

I wondered what 'tab was, and then vaguely remember I had
some trouble hear years ago and, thru trial-and error,
I found:

     (define-key c++-mode-map 'tab 'c-insert-my-tab)

worked.  However, I guess something has changed and now:

     (define-key c++-mode-map "\t" 'c-insert-my-tab)

doesn't produce the error.  I just have to figure out the
corresponding correction for:

  (define-key c++-mode-map 'iso-left-tab 'c-insert-my-shift-tab)

I remember searching some files in site-lisp for somewhere else
for the symbol:

  'tab
  'iso-left-tab

however, that was long ago and I can't remember how I found those
symbols.

Thanks for your efforts.

  -kind regards,
  Larry