Error report on startup

Error report on startup

[Herbert Euler]

Error report on startup could be confusing.  Consider the following in
~/.emacs:

    (with-current-buffer (find-file-noselect "some-file.el")
      (read (current-buffer)))

If ``some-file'' does not contain valid Lisp expression, the user will
see an error saying ``End of file during parsing:
/home/somebody/.emacs''.  However, the ~/.emacs file is valid; what is
not valid is the file some-file.el.

I had this problem when using Emacs on Windows.  I turned on
type-break mode.  For some reason I do not know until now, the file
~/.type-break was not written correctly for many times, so that the
functions `type-break-get-previous-time' and
`type-break-get-previous-count' failed on the invoking to `read'.
Then I saw the above message, this vague message took me so long to
find out the problem.

Is it possible to improve the error report?  An approach might be
setting `load-file-name' temporarily, but there seem to be many
adjustments.

Regards,
Guanpeng Xu

_________________________________________________________________
Don't just search. Find. Check out the new MSN Search! 
http://search.msn.com/

Re: Error report on startup

[Juanma Barranquero]

On 3/1/07, Herbert Euler <herberteuler@hotmail.com> wrote:

> Error report on startup could be confusing.  Consider the following in
> ~/.emacs:
>
>     (with-current-buffer (find-file-noselect "some-file.el")
>       (read (current-buffer)))
>
> If ``some-file'' does not contain valid Lisp expression, the user will
> see an error saying ``End of file during parsing:
> /home/somebody/.emacs''.  However, the ~/.emacs file is valid; what is
> not valid is the file some-file.el.

I don't see anything confusing. You're manually loading some-file.el
and reading its content in the context of loading .emacs, so certainly
it is ~/.emacs which is not valid (or, which is doing something
invalid). That the error comes from processing an elisp module is
irrelevant. It's not different to loading a non-lispy data file in
.emacs and then doing some kind of processing with it that causes an
error.

If instead you used `load' or `require' to load some-file.el you'd get:

  An error has occurred while loading `c:/home/.emacs':

  End of file during parsing: c:/emacs/site-lisp/some-file.el

  To ensure normal operation, you should investigate and remove the
  cause of the error in your initialization file.  Start Emacs with
  the `--debug-init' option to view a complete error backtrace.

             Juanma

Re: Error report on startup

[Herbert Euler]

>I don't see anything confusing. You're manually loading some-file.el
>and reading its content in the context of loading .emacs, so certainly
>it is ~/.emacs which is not valid (or, which is doing something
>invalid). That the error comes from processing an elisp module is
>irrelevant. It's not different to loading a non-lispy data file in
>.emacs and then doing some kind of processing with it that causes an
>error.
>
>If instead you used `load' or `require' to load some-file.el you'd get:
>
>  An error has occurred while loading `c:/home/.emacs':
>
>  End of file during parsing: c:/emacs/site-lisp/some-file.el
>
>  To ensure normal operation, you should investigate and remove the
>  cause of the error in your initialization file.  Start Emacs with
>  the `--debug-init' option to view a complete error backtrace.

It is just an explain to the general case.  If you are loading a library,
which you are not familiar with and it does this, how could you
know which file is invalid?

For me, the library is type-break mode, and the file is ~/.type-break.
If I did not read the source and did some digging work, I cannot
know it is the file ~/.type-break that causes the error.

Regards,
Guanpeng Xu

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Re: Error report on startup

[Juanma Barranquero]

On 3/1/07, Herbert Euler <herberteuler@hotmail.com> wrote:

> For me, the library is type-break mode, and the file is ~/.type-break.
> If I did not read the source and did some digging work, I cannot
> know it is the file ~/.type-break that causes the error.

Why are you loading type-break in such a weird way? If you .emacs contains

  (require 'type-break)

messages *will* be meaningful wrt the point of error.

             Juanma

Re: Error report on startup

[Herbert Euler]

>>For me, the library is type-break mode, and the file is ~/.type-break.
>>If I did not read the source and did some digging work, I cannot
>>know it is the file ~/.type-break that causes the error.
>
>Why are you loading type-break in such a weird way? If you .emacs contains
>
>  (require 'type-break)
>
>messages *will* be meaningful wrt the point of error.

Well, I load type-break in normal way.  Please take a look at the
functions `type-break-get-previous-time' and 
`type-break-get-previous-count'.
They contain similar expressions like I provided.  For example,
the functiion `type-break-get-previous-time':

(defun type-break-get-previous-time ()
  "Get previous break time from `type-break-file-name'.
Returns nil if the file is missing or if the time breaks with the
`current-time' format."
  (let ((file (type-break-choose-file)))
    (if file
        (timep ;; returns expected format, else nil
         (with-current-buffer (find-file-noselect file 'nowarn)
           (save-excursion
             (goto-char (point-min))
             (read (current-buffer))))))))

The file returned by `type-break-choose-file' is the actual
file of `some-file.el'.  On my system for some reason it is
not written correctly, which causes Emacs report confusing
error.

Regards,
Guanpeng Xu

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Re: Error report on startup

[Juanma Barranquero]

On 3/2/07, Herbert Euler <herberteuler@hotmail.com> wrote:

> (defun type-break-get-previous-time ()
>   "Get previous break time from `type-break-file-name'.
> Returns nil if the file is missing or if the time breaks with the
> `current-time' format."
>   (let ((file (type-break-choose-file)))
>     (if file
>         (timep ;; returns expected format, else nil
>          (with-current-buffer (find-file-noselect file 'nowarn)
>            (save-excursion
>              (goto-char (point-min))
>              (read (current-buffer))))))))

I stand by what I said. If type-break-get-previous-time loads elisp
code in such a way, it is to be *expected* that erroneous input (in
the file being read) will be reported as a problem of
type-break-get-previous-time. If the code's author would've wanted
normal loading semantics, s/he would've used (load file).

             Juanma

Re: Error report on startup

[Richard Stallman]

    I stand by what I said. If type-break-get-previous-time loads elisp
    code in such a way, it is to be *expected* that erroneous input (in
    the file being read) will be reported as a problem of
    type-break-get-previous-time. If the code's author would've wanted
    normal loading semantics, s/he would've used (load file).

It is not possible to do this by calling `load'.  `load' evaluates all
the sexps that it reads, and it throws the values away.

Re: Error report on startup

[Richard Stallman]

Does this patch give acceptable results?

*** type-break.el	30 Jan 2007 18:46:39 -0500	1.44
--- type-break.el	02 Mar 2007 14:33:25 -0500	
***************
*** 562,570 ****
      (if file
          (timep ;; returns expected format, else nil
           (with-current-buffer (find-file-noselect file 'nowarn)
!            (save-excursion
!              (goto-char (point-min))
!              (read (current-buffer))))))))
  
  (defun type-break-get-previous-count ()
    "Get previous keystroke count from `type-break-file-name'.
--- 562,573 ----
      (if file
          (timep ;; returns expected format, else nil
           (with-current-buffer (find-file-noselect file 'nowarn)
! 	   (condition-case nil
! 	       (save-excursion
! 		 (goto-char (point-min))
! 		 (read (current-buffer)))
! 	     (end-of-file
! 	      (error "End of file in `%s'" file))))))))
  
  (defun type-break-get-previous-count ()
    "Get previous keystroke count from `type-break-file-name'.

Re: Error report on startup

[Herbert Euler]

>Does this patch give acceptable results?
>
>*** type-break.el	30 Jan 2007 18:46:39 -0500	1.44
>--- type-break.el	02 Mar 2007 14:33:25 -0500
>***************
>*** 562,570 ****
>       (if file
>           (timep ;; returns expected format, else nil
>            (with-current-buffer (find-file-noselect file 'nowarn)
>!            (save-excursion
>!              (goto-char (point-min))
>!              (read (current-buffer))))))))
>
>   (defun type-break-get-previous-count ()
>     "Get previous keystroke count from `type-break-file-name'.
>--- 562,573 ----
>       (if file
>           (timep ;; returns expected format, else nil
>            (with-current-buffer (find-file-noselect file 'nowarn)
>! 	   (condition-case nil
>! 	       (save-excursion
>! 		 (goto-char (point-min))
>! 		 (read (current-buffer)))
>! 	     (end-of-file
>! 	      (error "End of file in `%s'" file))))))))
>
>   (defun type-break-get-previous-count ()
>     "Get previous keystroke count from `type-break-file-name'.

Sorry for the delay.  I could not use computer these days.  Yes,
it gives acceptable results.  However, there are actually two
`read's.  So the patch should be

*** type-break.el.orignal       Mon Mar  5 09:38:20 2007
--- type-break.el       Mon Mar  5 09:21:33 2007
***************
*** 562,570 ****
      (if file
          (timep ;; returns expected format, else nil
           (with-current-buffer (find-file-noselect file 'nowarn)
!            (save-excursion
!              (goto-char (point-min))
!              (read (current-buffer))))))))

  (defun type-break-get-previous-count ()
    "Get previous keystroke count from `type-break-file-name'.
--- 562,573 ----
      (if file
          (timep ;; returns expected format, else nil
           (with-current-buffer (find-file-noselect file 'nowarn)
!          (condition-case nil
!              (save-excursion
!                (goto-char (point-min))
!                (read (current-buffer)))
!            (end-of-file
!             (error "End of file in `%s'" file))))))))

  (defun type-break-get-previous-count ()
    "Get previous keystroke count from `type-break-file-name'.
***************
*** 576,585 ****
                (setq file
                      (with-current-buffer
                          (find-file-noselect file 'nowarn)
!                       (save-excursion
!                         (goto-char (point-min))
!                         (forward-line 1)
!                         (read (current-buffer)))))))
          file
        0)))

--- 579,591 ----
                (setq file
                      (with-current-buffer
                          (find-file-noselect file 'nowarn)
!                     (condition-case nil
!                         (save-excursion
!                           (goto-char (point-min))
!                           (forward-line 1)
!                           (read (current-buffer)))
!                       (end-of-file
!                        (error "End of file in `%s'" file)))))))
          file
        0)))

Thank you very much.

Best regards,
Guanpeng Xu

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Re: Error report on startup

[Richard Stallman]

    If ``some-file'' does not contain valid Lisp expression, the user will
    see an error saying ``End of file during parsing:
    /home/somebody/.emacs''.  However, the ~/.emacs file is valid; what is
    not valid is the file some-file.el.

I see your point, but is there a way we can do it better?

Which code in type-break.el causes this problem?
Many things visit .type-break, but I don't see anything that
evaluates it.