Table (used a "spreadsheet") org-sbe issues when the value is a string

Table (used a "spreadsheet") org-sbe issues when the value is a string

[Frederick Giasson]

Hi Everybody,

I am experiencing a few issues trying to update a cell value using 
=org-sbe= when the value of a column is a string.

Here is an example that works when the value is a number:


====================

#+NAME: average-dataset-size
#+BEGIN_SRC clojure :var f="turtle" :exports none :results value
(identity f)
#+END_SRC

#+RESULTS: average-dataset-size
: turtle


| Format | Sizes distribution in MB | Average size in MB |
|--------+--------------------------+--------------------|
|      7 |                          |                  7 |
#+TBLFM: $3='(org-sbe "average-dataset-size" (f $1))

====================

Now, if I put a string in the format column, I al getting the following:

====================

#+NAME: average-dataset-size
#+BEGIN_SRC clojure :var f="turtle" :exports none :results value
(identity f)
#+END_SRC

#+RESULTS: average-dataset-size
: turtle


| Format | Sizes distribution in MB | Average size in MB |
|--------+--------------------------+--------------------|
| turtle |                          | #ERROR             |
#+TBLFM: $3='(org-sbe "average-dataset-size" (f $1))

====================

The debugging session I have is (which doesn't tell me with Result: is 
#ERROR):

====================

Substitution history of formula
Orig:   ?
$xyz->  '(org-sbe "average-dataset-size" (f $1))
@r$c->  '(org-sbe "average-dataset-size" (f $1))
$1->    '(org-sbe "average-dataset-size" (f "turtle"))
Result: #ERROR
Format: NONE
Final:  #ERROR

====================


Note #1: (identity) is a function in Clojure that returns the same value 
that is in input.


I am wondering if this is a bug, or something that I am missing in how 
to use these features?


Thanks for helping,


Fred

Re: Table (used a "spreadsheet") org-sbe issues when the value is a string

[Nicolas Goaziou]

Hello,

Frederick Giasson <fred@fgiasson.com> writes:

> Now, if I put a string in the format column, I al getting the following:
>
> ====================
>
> #+NAME: average-dataset-size
> #+BEGIN_SRC clojure :var f="turtle" :exports none :results value
> (identity f) #+END_SRC
>
> #+RESULTS: average-dataset-size
> : turtle
>
>
> | Format | Sizes distribution in MB | Average size in MB |
> |--------+--------------------------+--------------------|
> | turtle |                          | #ERROR             |
> #+TBLFM: $3='(org-sbe "average-dataset-size" (f $1))
>
> ====================

According to `org-sbe' docstring, it should be

  #+TBLFM: $3='(org-sbe "average-dataset-size" (f $$1))

Documentation could be drastically improved in that area, the manual is
not even talking about `org-sbe'.

Regards,

-- 
Nicolas Goaziou

Re: Table (used a "spreadsheet") org-sbe issues when the value is a string

[Alex Bennée]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Hello,
>
> Frederick Giasson <fred@fgiasson.com> writes:
>
>> Now, if I put a string in the format column, I al getting the following:
>>
>> ====================
>>
>> #+NAME: average-dataset-size
>> #+BEGIN_SRC clojure :var f="turtle" :exports none :results value
>> (identity f) #+END_SRC
>>
>> #+RESULTS: average-dataset-size
>> : turtle
>>
>>
>> | Format | Sizes distribution in MB | Average size in MB |
>> |--------+--------------------------+--------------------|
>> | turtle |                          | #ERROR             |
>> #+TBLFM: $3='(org-sbe "average-dataset-size" (f $1))
>>
>> ====================
>
> According to `org-sbe' docstring, it should be
>
>   #+TBLFM: $3='(org-sbe "average-dataset-size" (f $$1))
>
> Documentation could be drastically improved in that area, the manual is
> not even talking about `org-sbe'.

Even this doesn't seem to work. In my case:

    #+name: scramble-string
    #+begin_src emacs-lisp :var str="abcde"
      (apply #'string
             (reverse
              (string-to-list str)))
    #+end_src

    #+RESULTS: scramble-string
    : edcba

So far so good

    #+call: scramble-string(str="whatever-foo")

    #+RESULTS:
    : oof-revetahw

So calling scramble-string works as expected...

    #+tblname: reversed-strings
    | abcde   | #ERROR |
    | flibble | #ERROR |
    | dibble  | #ERROR |
    | xyzzy   | #ERROR |
    | another | #ERROR |
    #+TBLFM: $2='(org-sbe scramble-string (str $$1))

Hmm no dice... the debugger lists:

    Substitution history of formula
    Orig:   '(org-sbe scramble-string (str $$1))
    $xyz->  '(org-sbe scramble-string (str $$1))
    @r$c->  '(org-sbe scramble-string (str $$1))
    $1->    '(org-sbe scramble-string (str $"abcde"))
    Result: #ERROR
    Format: NONE
    Final:  #ERROR

Even calling org-sbe directly from a source block:

    #+name: call-scramble-string-via-sbe
    #+begin_src emacs-lisp :var val="thing"
    (org-sbe scramble-string (str val))
    #+end_src

So is this just broken untested functionality or are there any working
examples from which we could reverse engineer the proper docstring?

--
Alex Bennée

Re: Table (used a "spreadsheet") org-sbe issues when the value is a string

[Nicolas Goaziou]

Hello,

Alex Bennée <alex.bennee@linaro.org> writes:

> So calling scramble-string works as expected...
>
>     #+tblname: reversed-strings
>     | abcde   | #ERROR |
>
>     | flibble | #ERROR |
>     | dibble  | #ERROR |
>     | xyzzy   | #ERROR |
>     | another | #ERROR |
>
>     #+TBLFM: $2='(org-sbe scramble-string (str $$1))

I cannot reproduce it. I get:

    #+name: reversed-strings
    | abcde   | edcba   |
    | flibble | elbbilf |
    | dibble  | elbbid  |
    | xyzzy   | yzzyx   |
    | another | rehtona |
    #+TBLFM: $2='(org-sbe scramble-string (str $$1))

> Even calling org-sbe directly from a source block:
>
>     #+name: call-scramble-string-via-sbe
>     #+begin_src emacs-lisp :var val="thing"
>
>     (org-sbe scramble-string (str val))
>     #+end_src

AFAIK, `org-sbe' is limited to tables.

Regards,

-- 
Nicolas Goaziou

Re: Table (used a "spreadsheet") org-sbe issues when the value is a string

[Alex Bennée]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Hello,
>
> Alex Bennée <alex.bennee@linaro.org> writes:
>
>> So calling scramble-string works as expected...
>>
>>     #+tblname: reversed-strings
>>     | abcde   | #ERROR |
>>
>>     | flibble | #ERROR |
>>     | dibble  | #ERROR |
>>     | xyzzy   | #ERROR |
>>     | another | #ERROR |
>>
>>     #+TBLFM: $2='(org-sbe scramble-string (str $$1))
>
> I cannot reproduce it. I get:

Hmm it turns out this interferes:

;; See http://emacs.stackexchange.com/questions/499/finding-and-executing-org-babel-snippets-programatically
(defun my-babel-hashed-confirm (lang body)
  "Check against known hashes before prompting for confirmation.
See `org-confirm-babel-evaluate'."
  (let ((check (list lang (md5 body))))
    ;; If not hashed, prompt
    (if (not (member check my-org-babel-hashes))
        ;; Ask if you want to hash
        (if (yes-or-no-p "Store hash for block? ")
            ;; Hash is added, proceed with evaluation
            (progn
              (add-to-list 'my-org-babel-hashes check)
              'nil)
          ;; Return 't to prompt for evaluation
          't)
      (message "Valid hash auto-confirmed")
      'nil)))

(setq org-confirm-babel-evaluate 'my-babel-hashed-confirm)

Resetting to org-confirm-babel-evaluate to nil and it works.

>
>     #+name: reversed-strings
>     | abcde   | edcba   |
>     | flibble | elbbilf |
>     | dibble  | elbbid  |
>     | xyzzy   | yzzyx   |
>     | another | rehtona |
>     #+TBLFM: $2='(org-sbe scramble-string (str $$1))
>
>> Even calling org-sbe directly from a source block:
>>
>>     #+name: call-scramble-string-via-sbe
>>     #+begin_src emacs-lisp :var val="thing"
>>
>>     (org-sbe scramble-string (str val))
>>     #+end_src
>
> AFAIK, `org-sbe' is limited to tables.

I was trying to replicate the way TBLFM compiles the macro.

>
> Regards,


--
Alex Bennée

Re: Table (used a "spreadsheet") org-sbe issues when the value is a string

[Alex Bennée]

Alex Bennée <alex.bennee@linaro.org> writes:

> Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:
>
>> Hello,
>>
>> Alex Bennée <alex.bennee@linaro.org> writes:
>>
>>> So calling scramble-string works as expected...
>>>
>>>     #+tblname: reversed-strings
>>>     | abcde   | #ERROR |
>>>
>>>     | flibble | #ERROR |
>>>     | dibble  | #ERROR |
>>>     | xyzzy   | #ERROR |
>>>     | another | #ERROR |
>>>
>>>     #+TBLFM: $2='(org-sbe scramble-string (str $$1))
>>
>> I cannot reproduce it. I get:
>
> Hmm it turns out this interferes:
>
> ;; See http://emacs.stackexchange.com/questions/499/finding-and-executing-org-babel-snippets-programatically
> (defun my-babel-hashed-confirm (lang body)
<snip>
>
> (setq org-confirm-babel-evaluate 'my-babel-hashed-confirm)
>
> Resetting to org-confirm-babel-evaluate to nil and it works.

So more correctly when org-confirm-babel-evaluate is set it breaks due
to info not containing a character position for the TBLFM code. I fixed
this by patching org-babel-check-confirm-evaluate:

modified   lisp/ob-core.el
@@ -242,7 +242,8 @@ should be asked whether to allow evaluation."
 		    (and export (equal eval "query-export"))
 		    (if (functionp org-confirm-babel-evaluate)
 			(save-excursion
-			  (goto-char (nth 5 info))
+			  (when (nth 5 info)
+			    (goto-char (nth 5 info)))
 			  (funcall org-confirm-babel-evaluate
 				   ;; language, code block body
 				   (nth 0 info) (nth 1 info)))

I'm not sure the goto-char is legitimate anyway as the documentation for
org-confirm-babel-evaluate says nothing about point being set while it
executes.

--
Alex Bennée

Re: Table (used a "spreadsheet") org-sbe issues when the value is a string

[Nicolas Goaziou]

Alex Bennée <alex.bennee@linaro.org> writes:

> So more correctly when org-confirm-babel-evaluate is set it breaks due
> to info not containing a character position for the TBLFM code. I fixed
> this by patching org-babel-check-confirm-evaluate:
>
> modified   lisp/ob-core.el
> @@ -242,7 +242,8 @@ should be asked whether to allow evaluation."
>  		    (and export (equal eval "query-export"))
>  		    (if (functionp org-confirm-babel-evaluate)
>  			(save-excursion
> -			  (goto-char (nth 5 info))
> +			  (when (nth 5 info)
> +			    (goto-char (nth 5 info)))
>  			  (funcall org-confirm-babel-evaluate
>  				   ;; language, code block body
>  				   (nth 0 info) (nth 1 info)))
>
> I'm not sure the goto-char is legitimate anyway as the documentation for
> org-confirm-babel-evaluate says nothing about point being set while it
> executes.

I agree and removed `goto-char'. Thank you for the analysis and the
patch.

Regards,