On Wed, 4 Oct 2006, Ned Freed wrote:
...
>> I don't quite like the part argument to "date", since the possible
>> values are listed explicitly. making each of them a tagged argument is
>> easier to express in the grammar, and it avoids the problem of
>> "${format}" which can fail during run-time. outline of suggested
>> change:
>
> The main reason I chose to make this a positional argument is that
> tagged arguments are properly optional and I don't think the type of
> date test should be allowed to be optional.
Tagged arguments != Optional arguments, despite looking the same. C.f.
section 2.6 of RFC 3028.
> This is quite unlike the sitations with other tagged arguments - they
> have sensible defaults.
It's like the :over/:user arguments to the 'size' test...which are tagged
and not optional.
Forgot about that one. It's one of things I like least in the base sieve
specification, so this strengthens my resolve to keep this as a positional
argument.
Ned