RE: 2119bis -- Tying our hands?
My first reaction was that the entire topic is a bike shed. The goal is clear and understandable specifications and 2119 is just a tool we use to make the process of producing and reading specifications more efficient. What I'm getting from this is that there are a significant number of drafts being submitted to the IESG for publication that are not sufficiently precise in their use of language. Deciding to revise 2119 seems - at first blush - an interesting reaction to this problem. But it's become evident that we don't share a common understanding of the language, nor are we consistent in its application. Formalizing usage patterns (c.f. server MUST/client MAY, MUST...UNLESS...) does help, if only because you've made it easier to do the right thing in more cases. If the problem is that authors and editors don't have the tools they need, then maybe a revision is the right approach. However, I'd still like to see more RFCs that describe something without resorting to using these crude implements. On 2011-08-31 at 05:08:15, Adam Roach wrote: > There is no reason to tie authors' hands by restricting them from > using perfectly good English words just because they happen to be the > same symbols used by RFC 2119. If we're going down this path, let's > scrap using MUST/SHOULD/MAY/etc, and formalize our conformance terms > with symbols that aren't English words. ___ Ietf mailing list Ietf@ietf.org https://www.ietf.org/mailman/listinfo/ietf
Re: 2119bis -- Tying our hands?
On 8/30/11 2:08 PM, Adam Roach wrote: Because the current suggestion -- which turns RFC writing into the game "Taboo" [1], but with incredibly common English words [2] as the forbidden list -- is ridiculous on its face. Don't use requirements language unless you absolutely have to. Otherwise, explain things in clear prose, describing what happens in normal and error cases, and the logic used in distinguishing them. If you absolutely require RFC 2119 requirements language to make something clear, I suggest the following symbology: ✔: MUST ☂: SHOULD ♥: MAY ✖: MUST NOT ♠: SHOULD NOT ☹: MAY NOT And the fact that the above is garbled for some large percentage of readers explains why we use 7-bit ASCII in drafts, so let's just stop that argument now, ok? -- Dean ___ Ietf mailing list Ietf@ietf.org https://www.ietf.org/mailman/listinfo/ietf
Re: 2119bis -- Tying our hands?
On 8/30/11 2:23 AM, Thomson, Martin wrote: On 2011-08-30 at 07:36:58, Peter Saint-Andre wrote: for long enough, I finally decided to submit an I-D that is intended to obsolete RFC 2119. IS THERE ANY CHANCE OF AGREEING THAT SHOUTING IS BAD? (i.e., Burger's first anti-law.) As opposed to mandating that requirements ought to be shouted. Lowercase "must" can be as effective as uppercase as long as it is consistently applied. You paint that as tongue-in-cheek, but Peter's draft does go down the rat-hole of picking out a color scheme for this particular bike shed, when doing so is really unwarranted. In addition to the SHOULD & co brouhaha, I have serious heartburn over this passage: When it is not appropriate to use the conformance terms, authors can use a variety of alternative words and phrases, such as: "need to" or "mandatory" instead of "MUST"; "ought to" or "strongly encouraged" instead of "SHOULD"; and "might" or "discretionary" instead of "MAY". To prevent confusion, authors ought to use these alternative words and phrases instead of the lowercase versions of the conformance terms, and to use the conformance terms only in their uppercase versions. There is no reason to tie authors' hands by restricting them from using perfectly good English words just because they happen to be the same symbols used by RFC 2119. If we're going down this path, let's scrap using MUST/SHOULD/MAY/etc, and formalize our conformance terms with symbols that aren't English words. Because the current suggestion -- which turns RFC writing into the game "Taboo" [1], but with incredibly common English words [2] as the forbidden list -- is ridiculous on its face. /a [1] http://en.wikipedia.org/wiki/Taboo_%28game%29 [2] According to Project Gutenberg, "must" and "may" are among the 100 words most frequently used in written literature. See http://en.wiktionary.org/wiki/Wiktionary:Frequency_lists#Project_Gutenberg ___ Ietf mailing list Ietf@ietf.org https://www.ietf.org/mailman/listinfo/ietf