Re: [Fwd: I-D ACTION:draft-freed-sieve-environment-ihave-00.txt]

["Mark E. Mallett" <mem@xxxxxxxxx>]

On Thu, Nov 16, 2006 at 08:38:33AM -0800, Aaron Stone wrote:
> 
> Let's say we dropped Ned's suggested requirement that actions must be
> used inside of ihave blocks testing for them.

That would be my quibble with 'ihave' as well.  While it might be
interesting to have a capability enabled for a block scope, I don't
think that doing this via a side-effect of a test inside of 'if' would
be the way to go about it.  It creates too much of a magical connection
between the evaluation of the test inside of the 'if' and the code block
that follows the 'if', and I think that would cause confusion to script
writers and Sieve language implementers, both.  It also asks for too
much context to be known inside the 'if' expression evaluator, e.g.:

   if not ihave "vacation" { ... }
   elsif something-else { ... # is vacation enabled here? }
   else { ... # how about here? }

the "ihave" has to know a lot about what's around it, like if there's a
'not' out there that, at least in my implementation, would actually be
applied after the 'ihave' test was already disposed of.  That's only a
very simple perversion, I can imagine others more complex :)  While I
could implement it, I wouldn't like to...

If one wants block scope of capabilities, I would think there are other
ways to accomplish it.  One way would be to have a new control command
with an attached block that could be else'd, e.g.:

    enable "vacation" {	... }
    else { ... }

Or simply separate the test and the enable, e.g.:

    if ihave ["vacation", "notify" ] {
        enable ["vacation", "notify"];
	   ...
    }

If desired, this could be defined so that the effect of the enable went
away when its containing block ended.  I dunno if I would want that, I
guess I'd have to think about it; I tend to prefer more explicit
twiddling (e.g. by also having a complementary "unenable").  Then again,
I don't really see why one would want the capability to go away once
enabled.

Also I wonder about this bit:

   > Ihave is designed to be used with extensions that add tests, actions,
   > or comparators.  It MUST NOT be used with extensions that change how
   > the content of Sieve scripts are interpreted such as the variables
   > extension [I-D.ietf-sieve-variables]

Why?

mm