[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Must and should: new proposal (was: Re: Must and should again)

On Wed, Apr 18, 2001 at 12:35:46AM +1000, Anthony Towns wrote:
> aj, who'd rather relying on things that are objectively verifiable, rather
>     than oracles like the policy editor or the release manager

The RFC usages of SHOULD and MUST have spread far beyond the RFCs,
they are popular among groups that write standards and protocol
descriptions.  I've seen standards from widely different sources,
and if they define "should" and "must" as technical terms, they
always refer to the RFC usage or define something similar.
(They then tend to go ahead and use them wrong, but still.)

Defining "should" and "must" to something different is objectively
bad style, because it breaks established conventions in an unexpected
way.  It's a bit like writing C code that uses "i" as a random pointer
and "p" as a loop index.  If you're coding for the IOCCC you can even
keep writing p[i] :-)

You can expect people to go read the intro that defines should and
must, when they read parts of policy in isolation, but I think you
will be disappointed in that expectation.  Keeping the difference
in mind also uses up valuable neurons that could be used to think
up better ways to write maintainer scripts.

Richard Braakman

Reply to: