Hi Ihor, Ihor Radchenko <yanta...@posteo.net> writes: > For example, consider > > #+begin_src bash > mkdir foo > #+end_src > > #+begin_src emacs-lisp > (write-file "foo/result") > #+end_src > > The second block relies upon side effects of the first block. > However, if ob-shell is not loaded, with your patch, the side effect > will not happen. One can construct more nasty examples when not > executing prior block leads to unexpected or even damaging results.
If I understand correctly, you are saying that it is better to let the command crash as soon as possible instead of letting it execute potentially problematic code because some previous block has not been executed. I think it makes perfect sense. I'm now convinced that my patch is not desirable. Feel free to close the issue. > Why not simply adding :eval no header argument? Just because I didn't know it exists. Thank you for telling me. I have now added it to all my code blocks generating non-executable results using the :wrap keyword like this: #+name: show-tree #+begin_src sh :exports both :results output drawer :tangle yes :wrap "SRC text :eval no" echo foo #+end_src #+RESULTS: show-tree #+begin_SRC text :eval no foo #+end_SRC It works very well, thanks! Would you say this is how you would have done it as well or is there a better way? -- Damien Cassou "Success is the ability to go from one failure to another without losing enthusiasm." --Winston Churchill