Thanks for the investigation Ian.  So, since the tests run just fine
without it, and it offers an inconsistent and at times detrimental
feature, can we consider removing it, and/or adding some options for
that?

I'd be fine having to flag my src-block with a ":verbatim t" option to
make sure that the output is not mangled.

Thoughts?

On Fri, Jul 17, 2020 at 7:30 AM ian martins <ianxm@jhu.edu> wrote:
>
> Fortunately the author wrote tests, so we can tie the behavior of the code to use cases. Unfortunately all the tests pass with the call to org-trim removed. Also the call is there from the first commit of the file in git, so there's no commit message to explain.
>
> My guess is that it was added to clean up cases that resulted in extra trailing whitespace, but I dunno.
>
>
> On Wed, Jul 15, 2020 at 7:12 PM Michaël Cadilhac <michael@cadilhac.name> wrote:
>>
>> Hello,
>>
>> Quick question here: in ob-C.el, before returning the output of a C
>> file, there's this line:
>>
>> (setq results (org-trim (org-remove-indentation results)))
>>
>> That seems quite arbitrary; is it on purpose?  I have a C file that
>> outputs some sort of list of formatted numbers, e.g.:
>>
>>   0  -17.8
>>  40    4.4
>>  80   26.7
>> 120   48.9
>>
>> and only the first line gets trimmed, leading to a faulty output.
>>
>> This does not seem to be a universal thing in Babel; for instance:
>>
>> #+begin_src emacs-lisp :exports both :results value raw
>>   " 0\n 1\n2\n"
>> #+end_src
>>
>> …results in:
>>
>> #+RESULTS:
>>  0
>>  1
>> 2
>>
>> But the same thing in C:
>>
>> #+begin_src C :exports both :results output raw
>>   printf (" 0\n 1\n2\n");
>> #+end_src
>>
>> …results in:
>> #+RESULTS:
>> 0
>>  1
>> 2
>>
>> Cheers,
>> M.
>>