On Fri, 14 Nov 2014, Nicolas Goaziou wrote:
"Charles C. Berry" <ccbe...@ucsd.edu> writes:
More patches (as you can see). Now ox.el, ob-core.el, and ob-exp.el
are patched.
Thanks.
[skipping to the bottom - omitting useful critiques of code and
opinions about strategy and tactics from Nicolas]
WDYT?
After staring at `org-babel-insert-result' for too long, I am beginning to
feel like Alice in Wonderland. As currently implemented, inline src blocks
are somewhere between fragile and broken. I worry about making them even
more fragile, the logic in `org-babel-insert-result' has plenty of twists
ans turns, and I cannot commit the effort to dig deeply into them. So I am
looking for an easy way out.
For now, I'd be willing to make patches that will allow removal of the
inline src block results that do *not* involve these header args:
- :file <fn>
- :wrap <wrapper>
- :results latex html drawer org code
which I can do barely touching `org-babel-insert-result' and this
simplifies matters a lot.
I propose to do this by using the patches of ox.el and ob-exp.el from
my last post. For ob-core.el, I would
- leave defcustom org-babel-inline-wrap as "=%s=" (or use defconst - I
do not have a strong opinion either way).
- allow replacement of exising results in `org-babel-insert-result'
like so
#+BEGIN_EXAMPLE
- (existing-result (unless inlinep
-(org-babel-where-is-src-block-result
+ (existing-result (if inlinep
+(org-babel-delete-results-macro)
+ (org-babel-where-is-src-block-result
#+END_EXAMPLE
and defun `org-babel-delete-results-macro' per Nicolas' suggestions
for `org-babel-delete-babel-snippet' (from earlier patch) to use
`org-element-context' and friends (and not mess with export
snippets).
- modify `org-babel-examplify-region' along these lines
#+BEGIN_SRC emacs-lisp
(insert
(replace-regexp-in-string
"," "\\,"
(format
(concat "{{{results("
org-babel-inline-result-wrap
")}}}"
(prog1 (buffer-substring beg end)
(delete-region beg end))))
nil t))
#+END_SRC
I believe that this is simple enough to avoid breaking idioms that
folks might use now.
As for the choice between "=%s=" and "%s", the latter was hard coded
until f285b7ed3d097dd1cbb55fa3c31bc92aa0149054 in February 2013 and
has been the default since. It also parallels what happens with
handling src block results. Going forward I do not think this
behavior should change.
I have too litle experience with #+MACROs to know if Aaron's
suggestion to let the user customize the macro is opening up potential
issues when users get `creative'.
I can do what I've outlined in the coming days.
If it is felt that more retooling of `org-babel-insert-results' is really
needed, I can get to it early next year. In fact, I'll be out of email
range from late this month till then, so any problems I create now will
have to wait till then for me to work on them.
Thanks for the critique of my earlier patches and your thoughts.
Best,
Chuck
