Bruce Lilly wrote:
Looking towards the future, ideally we won't be forced by today's
decisions into a situation where there are once again umpteen
widely-scattered documents that amend the basic IMF format in
various ways. One way to avoid that undesirable scenario is to
make at least the parse grammar as liberal as practical in
accordance with the robustness principle.
Looking towards the future I'd pick "net-utf8" as a least common
denominator for line oriented data (not binary or XML). Net-utf8
would move any HTAB to "obs" (unless John changes this detail),
and move all CTL apart from CRLF, LF, HTAB, and FF to /dev/null.
Frank