[rfc-i] Incorrect use of the word "ASCII" in section 3.2

Joe Hildebrand (jhildebr) jhildebr at cisco.com
Mon Mar 11 13:40:09 PDT 2013

On 3/10/13 1:51 PM, "Paul Hoffman" <paul.hoffman at vpnc.org> wrote:

>My best guess is that what is meant for "ASCII" is "the characters U+0021
>through U+007E" and for "non-ASCII" it is "characters other than U+0021
>through U+007E". If that guess is correct, the above paragraph should

I bet you meant to include U+0009 (tab), U+000A (line feed), U+000C (form
feed), and U+000D (carriage return).

>The official language of the RFC Series is English.  All text that must
>be read to understand or implement the technology described in the RFC
>must use characters U+0021 through U+007E.  Use of characters outside
>that range must receive explicit approval from the document stream
>manager and will be allowed after the rules for the common use cases are
>defined in the Style Guide.

I mostly agree with that, with a further expository note that says that
there WILL be some cases (such as examples) where characters outside of
US-ASCII will be allowed, so downstream formats MUST be prepared to handle
them in some manner, even if that manner is not pleasant to read.

>Note that this proposed wording does not include anything about a Unicode
>Encoding Form. Some Publication formats might not use *any* encoding
>form, and trying to wind that into this paragraph causes more confusion.

Agree.  I would argue that the canonical format should only ever use
UTF-8, but that's a separate point.

>P.S. Please do not interpret my offering of technically more correct
>wording above as supporting the idea in the paragraph. I think it is
>unnecessarily restrictive to prevent the use of ASCII in text that is
>needed to be read or understood. It's 2013, folks. If an example that is
>needed to be understood contains a character such as ü, it is better to
>include the character directly than putting in some placeholder and
>trying to describe that in the text surrounding the example. I hope ADs
>in the future can convince the RSE to change this rule for IETF Stream

I also agree with this.  I'd like to leave the issue of where codepoints >
U+007F can be used completely up to the standards of future generations,
encoded in the style guide, up for renegotiation at any point that makes
sense according to community input with the RSE's expertise as final

Joe Hildebrand

More information about the rfc-interest mailing list