Nico,

> At some point, someone will want to write foo#call_bar() without
> #call_bar() being replaced, and we will go back to point 0. IMO,
> zero-width space is the way to go, if only because it can already be
> used to escape other special characters in Org.

oof, of course!  color me convinced!

one concern: if i essentially "revert" the changes in
7efa0f2879226695ad9b309c9263a4d1b5d79e89, call_ and src_ all work, but i
get the ZERO WIDTH SPACE (ZSWP, \u200b) propagated into the resulting
output file (.html, .tex). it doesn't seem to do any "harm", but, i find
it disquieting (being an ASCII-type by birth, if not by current usage).

the below patch, basically my #-paste patch turned into a \u200b-paste
patch, gets rid of the ZWSP in the output file in the case of
\u200bcall_, \u200bsrc_ (**), but, leaves it for all other uses, i.e.,
\u200b=fubar=.  this inconsistency i *also* find disquieting, sigh; i
can imagine the great-regexp in the sky in org-element.el could get rid
of all these prefixing ZWSPs, but i'd not proceed in this direction
without (policy, at least) direction.

cheers, Greg

(**) btw, if one wants \u200b<<RESULTS OF CALL_FOO()>> in the output,
one prefixes CALL_FOO() with *two* ZWSPs.