On Tue, Nov 08, 2005 at 08:23:27PM +0100, Michael Nordstrom wrote:
>> 2) The word "should" effectively means "may not" and thus creates a
>> loophole for misinterpretation, and so we should use "must"
>> (yes, it's a bad pun). ;-)
> Well, I agree that "should" wasn't the correct word, but not for the
> reason you list. I think we should use the kind of terminology that
> is used in standards (e.g by IEEE) and in that kind of documents
> "should" equals "is recommended that"; I should have used "shall"
> instead ("shall" equals "is required to"). The word "must" is
> deprecated in standards.
Actually, I kept RFC 2119 in mind.
But I see no point to agrue on this: if you feel like using
"shall", go on and use it. I fully agree.
BTW, If you may point to documents standardizing such wording (I
mean those of IEEE, you've mentioned, for example), I think it's
worth including references to them into preamble of the Plucker DB
format document. It's the common practice with RFCs.
>> "Any sequence of continued text records must be treated as a single
[...]
>> them since they derive it from the first record in a sequence".
> It is a very good suggestion as long as we replace "must" with
> "shall".
So let's stick to it. With a typo fixed and "shall" applied it goes
like this:
"Any sequence of continued text records shall be treated as a
single solid record from the standpoint of its properties. If a
sequence of continued text records is to have any properties that
may be attached to a non-continued text record (i.e. navigational
metadata, exceptional charset, etc) such properties shall be
attached to the very first record in such a sequence; all other
records in a sequence shall not have such properties attached to
them since they derive these properties from the first record in a
sequence".
_______________________________________________
plucker-dev mailing list
[email protected]
http://lists.rubberchicken.org/mailman/listinfo/plucker-dev
