On Wed, May 9, 2018 at 3:48 PM, Jean-Christophe Helary <brandelune@gmail.com> wrote:
You misquoted me. I was talking about design constraints when C and Lisp were created, which kept language creators from "inventing" proper language localization. I was specifically replying to Diego Zamboni regarding that.

I don't think it was only those constraints. Imagine if C and LISP had been designed with "keywords in your own language" in mind. I'm pretty surre that would have largely impeded the proliferation of compilers/interpreters that made possible the explosion of those, and many other, languages.

I fully agree with Nicolas that in this context, localization should be a display problem and not involve modifying the source. Take for example the educational language Scratch (https://scratch.mit.edu/), in which you can localize the language (i.e. the blocks with which you build your programs). However, if you download the source for your program (it's a JSON file), it's always the same, no matter in which language you have the interface.

As a first step, you can already configure Emacs so that the markup is minimally visible. Look at this screenshot, for example: http://zzamboni.org/post/beautifying-org-mode-in-emacs/emacs-init-propfonts.png. Most of the formatting is visually communicated, you can only see a few keywords (properties, begin_src, etc.). It really is very non-intrusive in my opinion.

--Diego