On Thu, 12 Jun 2014, Eric Schulte wrote:
Charles Berry <ccbe...@ucsd.edu> writes:
The following
,----
| #+PROPERTY: tangle yes
|
| #+NAME: print-abc
| #+BEGIN_SRC emacs-lisp
| (format
| "<<%s>>" "abc")
| #+END_SRC
|
| #+NAME: print-def
| #+BEGIN_SRC emacs-lisp
| (concat
| "<<" "def" ">>")
| #+END_SRC
|
| #+BEGIN_SRC emacs-lisp :noweb yes
| <<print-abc>>
| <<print-def>>
| #+END_SRC
`----
tangles as
,----
| (format
| "<<%s>>" "abc")
|
| (concat
| "<<" "def" ">>")
|
| (format
| "<<%s>>" "abc")
| (concat
| "")
`----
Note that ""<<" "def" ">>" disappeared from (concat ...).
FWIW, adding quotes inside the constant string in org-babel-noweb-wrap:
"\\([^ \"\t\n][^\"]+?[^ \"\t]\\|[^ \"\t\n]\\)"
fixes this instance.
Does this raise any error, or fail silently?
Thanks,
Eric
Alas, it fails silently -- producing the erroneous output.
I guess the issue is that
`(re-search-forward (org-babel-noweb-wrap) nil t)'
finds quoted expressions like
1. "<<abc>>"
2. (concat "<<" abc ">>")
FWIW,
(re-search-forward
(org-babel-noweb-wrap
"\\([^ \"\t\n][^\"]+?[^ \"\t]\\|[^ \"\t\n]\\)"
) nil t)
(adding quotes to the [^...] bits) takes care of example 2. Not sure
if example 1 is a feature or a bug - I sometimes use things like
"<<text-I-want>>"
so maybe modifying the regex is good enough?
OTOH, <<no-such-chunk>> has no effect. Maybe checking that the reference
actually resolves is better than tweaking the regexes.
HTH,
Chuck
