Re: [Toss-devel] New Speagram: plan of action

[Lukasz S. <luk...@gm...>]

On Wed, Jun 20, 2012 at 7:32 PM, Lukasz Kaiser <luk...@gm...> wrote:
> Hi.
>
>> The spec that is posted to the list.
>
> I get that - I was not sure which part exactly. Because in the one
> that I was thinking about, it seems to me that the definition of
> terms was exactly as our OCaml term type - no need to update.
> So what am I missing?

I've come to the conclusion that a variable can have (as originally)
only one type, which is different from a term which can have several
supertypes.

>>> Sure - that will hopefully disambiguate all casts!
>>
>> But this step is is just about implementing the algorithms.
>
> I am very much against implementing algorithms without
> adjusting the tests and core language at the same step.
> (By "step" I mean something bigger than a commit, of course.)
> And when they are adjusted, a lot of ambiguities will be gone :).

I don't agree. I'm talking about basic algorithms (which don't have
"wiggle room"), not auxiliary algorithms like adding an intermediate
node in inheritance hierarchy stored as a well-formed set of
declarations.