🚀 go-pugleaf

fjh@munta.cs.mu.OZ.AU (Fergus Henderson) wrote:
>Paul Long <plong@perf.com> writes:
>>A language standard is not a
>>set of suggested guidelines for tool developers--it is a very exact,
>>necessarily legalistic contract between the consumers and producers of these
>>tools. As consumers, we should settle for no less.
>
>I think a language standard should be exact (and legalistic) where possible,
>but there are many situations where for various reasons this is not
>possible, and in these situations I think a language standard _should_
>offer guidelines.

If the writers of a standards document wish to influence an implementation
without requiring specific behavior, they need to make this distinction
absolutely clear. Ways to accomplish this: place commentary in special
sections, such as an annex, in a separate document, or in non-binding
footnotes, or use special language to indicate desirable behavior. In
telecommunications standards documents, I've seen this distinction made with
"shall" and "must" indicating a requirement, while "should" and "it is desired
that" indicating a, well, desire.

(BTW, I believe that someone said that the ANSI footnotes were not binding, but
that the same footnotes in the ISO document are binding. Can anybody comment on
this?)

>The Ada standard does this well, with "Implementation Advice" sections;
>the C standard is not nearly as good in this respect, IMHO.

Haven't seen it.

-- 
Paul Long                               45:29:14N 122:48:09W
plong@perf.com                          http://www.teleport.com/~pciwww/
    "Where time or intelligence are lacking, a goto may do the job."
        - M.E. Hopkins, "A Case for the GOTO," 1972.