On 2015-05-28 Thu 11:15, Thomas S. Dye wrote:
> Titus von der Malsburg <malsburg@posteo.de> writes:
>
>> I can’t reproduce the second indent but I think it’s a bug that there is
>> any indentation at all.
>>
>> The documentation of `org-edit-special' (C-x ') says:
>>
>>   Call a special editor for the element at point. …
>>
>> No mention of indentation or other reformatting of my code.
>>
>> The same goes for `org-edit-src-exit' (C-c C-c) which says:
>>
>>   Kill current sub-editing buffer and return to source buffer.
>>
>> The edit-in-buffer feature should not touch the indentation.  If the
>> syntax of the language is sensitive to indentation (e.g. Python) this
>> can break the code.  Example:
>>
>> #+BEGIN_SRC python :results output
>>   print "test"
>> #+END_SRC
>>
>> is invalid Python syntax.
>>
>> Also having one function perform two very different actions (edit code
>> in separate buffer *and* reformat the code) is poor design.  At least in
>> this special case.  When I open the code in a separate buffer but then
>> decide not to change it (C-c C-c), I'll end up with extra indentation
>> and this will create unnecessary changes when I commit the file in git.
>
> Does the variable org-src-preserve-indentation get the behavior you're
> after?

It does.  I propose two changes:

1.) The default value `org-src-preserve-indentation' should be t because
that is safer (see Python example) and more conservative.  (Don’t mess
with my code!)

2.) The documentation of `org-edit-special' and `org-edit-src-exit'
should point out that depending on `org-src-preserve-indentation' the
code may be re-indented.

Even better would be to remove `org-src-preserve-indentation' and to do
the right thing, which is to leave the code untouched.  Emacs already
has facilities for indenting code.  No need to reinvent the wheel.

  Titus