Re: How to override ":eval no" in call lines?

["Charles C. Berry" <ccberry@ucsd.edu>]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 3387 bytes --]

On Mon, 9 Feb 2015, Sebastien Vauban wrote:

> "Charles C. Berry" wrote:
>> On Fri, 23 Jan 2015, Sebastien Vauban wrote:
>>> "Charles C. Berry" wrote:
>>>> Sebastien Vauban wrote:
>>>>> In a long document, I must have ":eval no" at file level, as this
>>>>> is the common setting for most code blocks. However, how do I unset
>>>>> that for some call lines.
>>>
>>> I don't get why one has to add ":eval yes" for both types of headers
>>> arguments.
>
> I still don't get that: why do I need to add *twice* ":eval yes", in
> both the "inside header args" and the "end header args"?
>
> The documentation [1] states:
>
>  ┌────
>  │ END HEADER ARGUMENTS are applied to the calling instance and DO NOT
>  │ AFFECT EVALUATION OF THE NAMED CODE BLOCK. They affect how the
>  │ results are incorporated into the Org mode buffer and how the call
>  │ line is exported.
>  └────
>
> If end header args don't affect the evaluation of the name code block,
> why do we have to set ":eval" to "yes", then?
>

Because there are two evaluations to be made of a call_abc() instance or a
`#BEGIN_SRC lang :var x=abc() ...' instance:

1. one of abc()

2. one of the instance.

They can be made in any of the four combinations of `:eval yes' and `:eval 
no'. See below for an example of a src block calling another using the 
`:var x=abc()' idiom.


>>> Moreover, I once read that when evaluating a call line, it is
>>> converted into an ephemeral Emacs Lisp code block equivalent to the
>>> call line (and created at the point of the call line):
>>>
>>>  #+begin_src emacs-lisp :var result=<NAME>(<ARGUMENTS>) <INSIDE-HEADER-ARGS>
>>>    result
>>>  #+end_src
>>>
>>> which is evaluated in place.
>>
>> No, like this:
>>
>> #+begin_src emacs-lisp :var result=<NAME>[<INSIDE-HEADER-ARGS>](<ARGUMENTS>)
>                                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>
> What's that syntax?  The one described for "header arguments in function
> calls"?  Aren't we recursive here: describing syntax equivalent to
> a call via the ephemeral code block, reusing syntax for a call?
>

Not sure how best to answer. Maybe try out all combos to demo what
happens:

A simple src block:

#+NAME: up
#+BEGIN_SRC emacs-lisp :var x="CbA" 
(upcase x)
#+END_SRC


Eval this src block and `up' - prompted twice for evalution upon
execution.  Note RESULTS == 'C B A':

#+BEGIN_SRC emacs-lisp :eval yes :var x=up[:eval yes](x="c b a") 
x
#+END_SRC

#+RESULTS:
: C B A

Do not eval this src block but eval `up' - prompted to evaluate `up'
and message that evaluation is disabled appears for current src
block. No RESULTS:

#+BEGIN_SRC emacs-lisp :eval no :var x=up[:eval yes](x="c b a") 
x
#+END_SRC


Eval this src block and not `up' - prompted once for evaluation.  Note
RESULTS == 'nil', as x did not get a value assigned to it:


#+BEGIN_SRC emacs-lisp :eval yes :var x=up[:eval no](x="c b a") 
x
#+END_SRC

#+RESULTS:
: nil

Eval neither. No prompts. No result.

#+BEGIN_SRC emacs-lisp :eval no :var x=up[:eval no](x="c b a") 
x
#+END_SRC


>>> Where do <END-HEADER-ARGS> fit into that picture?
>>
>> Either before or after the :var ...
>>

Maybe better to say:

In this context the equivalent of <END-HEADER-ARGS> is the `:eval'
header for the src block, which can go anywhere on the line. The
<END-HEADER-ARGS> if supplied in this context seem to be ignored.

HTH,

Chuck