[rfc-i] draft-rfc-image-files-03

Peter Saint-Andre stpeter at stpeter.im
Tue Apr 10 17:14:45 PDT 2012

On 4/10/12 4:40 PM, Paul E. Jones wrote:
> Julian, et al,
>>> Not specific to this HTML example, but one thing we night want to
>>> consider is whether we want to change the normative words from all
>>> uppercase to italics (or bold) lowercase.  This document shows
>>> uppercase, because that is the current agreed style driven by the use
>>> of ASCII.  HTML opens up an opportunity to change that, if we want.
>> I'm currently using all uppercase with a different font because that way,
>> copying and pasting to plain text will do the right thing.
> Does this suggest a different meaning:
>   "The client MUST select..."
>   "The client must select..."
> I can appreciate emphasizing words (for a quick scan through a document),
> though I wonder if use of certain tags could address that need.  For
> example, I can load an HTML document in Word that has normative words (in
> whatever case) tagged with class="rfc2119" and ask Word to highlight all of
> them.

I'd rather not rely on HTML formatting, because it's quite natural for
people to copy-and-paste text from the original context into plaintext
email messages, code comments, etc. Much as I don't like shouting, I
think the uppercase conformance terms are here to stay.

> So, is there really a need with modern browsers, editors, etc. to even
> provide emphasis?  I can appreciate how it would be useful for a quick
> eye-scan over a document, but I would not think it would generally be useful
> when carefully reading a spec... unless "must" and "MUST" really means
> something different.

In my practice of writing specs, I now scrupulously avoid the lowercase
equivalents of the uppercase conformance terms. It's easy enough to use
"needs to" or "has to" instead of "must", "ought to" or "is encouraged
to" instead of "should", "might" or "can" instead of "may", etc.


Peter Saint-Andre

More information about the rfc-interest mailing list