Arnt Gulbrandsen wrote:
I've used a number of systems that use TAB instead of SP;
they're still out there and should not be considered invalid.
I am persuaded by this argument.
Charles proposed SHOULD, without SHOULD it's just an advice,
nothing remotely related to "invalid".
[you have]
sent umpty million colon-tab messages. It attracted more than
its fair share of problems, and I wouldn't recommend doing it.
In other words: If implementors go to the trouble to insert a
WSP after the colon (excluding cases of folding immediately
after the colon, and ignoring some simple minded scripts) they
better pick SP instead of HTAB for this WSP.
Your argument supports Charles' proposal to mention this, the
current net-utf8 draft also "dislikes" HTAB-incarnations of WSP.
I was unhappy with this detail precisely because of WSP in mail,
but "accept HTAB, better don't create it" makes a lot of sense
for the discussed "after colon" position. For "after CRLF" in
a folded header field I'd be more stubborn in defending HTAB :-)
Frank