Re: Alternate language spec without declare-{input,output,parameter}

I think these are fixable:

Henry S. Thompson wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> This is broken as it stands, wrt parameters, and possibly inputs.
> 
>  1) We've lost the ability to make parameters obligatory/optional,
>     which I'm sure we agreed on ages ago -- clearly this is only
>     allowed when declaring a parameter;

Yes.  See next comment.
> 
>  2) It's incoherent to use a wildcard (* (or *:x, or x:* for
>     parameters) as the name of an actual input or parameter.  Those
>     only make sense as part of signatures, that is, when the
>     input/parameter is being declared. . .

In the case of declaring parameters in component types, we need the
wildcards and the ability to say whether they are required.

In the same way in which alternatives are shown in the document, I think
we could a bit that says:

   "When declaring a parameter in ... the following syntax is available: 
..."

--Alex Milowski

Received on Thursday, 26 October 2006 15:06:22 UTC