Hi Sebastien

"Sebastien Vauban"
<wxhgmqzgw...@spammotel.com> writes:

> Hi Eric,
>
> Eric Schulte wrote:
>> "Sebastien Vauban" <wxhgmqzgw...@spammotel.com> writes:
>>> If there is no value assigned to the code block...
>>> ... there is no an error when exporting:
>>>
>>> #+begin_src text
>>> Debugger entered--Lisp error: (wrong-type-argument consp nil)
>>>   setcdr(nil "nil=x")
>>> #+end_src
>>
>> All variables must be given a default value, which is why it is an error
>> to write a code block like the following.
>>
>> #+source: circle(x)
>> #+begin_src python
>> return x*x
>> #+end_src
>>
>> I've just pushed up a change which provides a better error message in
>> this case.
>
> It works OK. Thanks a lot for that.
>
> Remark: This did correct a problem that I had with my local LOB, and which
> appeared a couple of months ago. I tried to find the post back, but I
> couldn't. At that time, you tried my example, but you couldn't reproduce it.
> Anyway, I now could check that it was because of this, now that the message is
> a lot clearer than "consp, nil".
>
> Question: Would it be possible to add the src-name in the error message?  You
> know, in my LOB, I have so many times the same var name coming back again,
> that having the name of the source block would be of a great help. I tried
> looking into the code to see if I could do it myself, but I must admit it's
> still too hard for me. I have the impression many functions must be extended
> in order to pass the src-name down to where the error is thrown...
>
> Other thing, I tried to add a test for checking your fix never will be
> regressed. This is what I currently have:
>
> * Does not work
>   :PROPERTIES:
>   :ID:       f2df5ba6-75fa-4e6b-8441-65ed84963627
>   :END:
>
> If there is no value assigned to the code block, a proper error should be
> thrown.
>
> #+source: carre(x)
> #+begin_src python
> return x*x
> #+end_src
>
> * Test
>
> #+begin_src emacs-lisp
> (ert-deftest test-org-babel/no-defaut-value-for-var ()
>   "Test that the absence of a default value for a variable does throw a proper
>   error."
>   (org-test-at-id "f2df5ba6-75fa-4e6b-8441-65ed84963627"
>     (org-babel-next-src-block)
>     (should-error (org-babel-execute-src-block))
>     :type 'error))
> #+end_src
>
> Though, I have 2 questions:
>
> - How can I differentiate between the clean error (with a message) and the one
>   which wasn't correctly trapped?  Based on the first line of a backtrace
>   (string comparison) or on the type of the error?  In the latter case, how
>   can I know what's the type of the current error thrown, and the one of the
>   error before your fix?

I can help with the error question. You can test the specific error
message as follows:

--8<---------------cut here---------------start------------->8---
(ert-deftest test-org-babel/no-defaut-value-for-var ()
  (org-test-at-id "f2df5ba6-75fa-4e6b-8441-65ed84963627"
    (org-babel-next-src-block)
    (let ((err
           (should-error (org-babel-execute-src-block)
                         :type 'error)))
      (should (equal '(error "variable \"x\" must be assigned a default value") 
err)))))
--8<---------------cut here---------------end--------------->8---
 
>
> - I wonder why we need twice the =org-babel-next-src-block= call, and not only
>   once in the =should-error= form.
>
> Thanks.
>
> Best regards,
>   Seb

Best, Martyn


Reply via email to