On Tue, 26 Aug 1997, Jacob Palme wrote:
> At 11.25 +0100 97-08-26, [log in to unmask] wrote:
> > RFC 1738 describes how to encode URLs. Clearly, if we only encounter URLs
> > in text/html objects that are RFC 1738 compliant we have no problem as they
> > will also be RFC 822 header syntax compliant. The problem arises when we
> > encounter URLs that are not RFC 1738 compliant and which, therefore, may
> > not be RFC 822 header syntax compliant. This then is the case, and the only
> > case, in which 2047 encoding is required.
> That is a little vague to write in a standard. I would prefer to say that
> all URLs which are not RFC 1738 compliant MUST be encoded with RFC 2047
> encoding. That is, that we should not specify our own definition of which
> character codes need and need not 2047 encoding.
Well, I have no problem agreeing. But either way, we would not create our
own definition. In one case, we would use the 822 definition, in the
other case the 1738 definition. From an implementation viewpoint
(I'm not an implementer, just guessing), I assume that basing it on
822 is easier, because the average MIME-aware mail software already
has a function: Encode-this-header-if-necessary (where necessary
is defined in 822 terms).