Generic parameter of StructuredTaskScope

[Alan Bateman]

On 12/05/2023 21:11, Attila Kelemen wrote:
> Hi,
>
> There is an inconsistency in the treatment of the generic parameter of
> STS. Namely, that while it was deemed important to know the common
> base type of the result of the forked tasks; Surprisingly, its two
> common implementations do not provide this feature. How come that in
> the STS implementations believed to be used almost always are not
> believed to find this feature useful? I mean, in the use cases I can
> think of relying on this generic parameter, I would happily rely on it
> even in the default implementation.
The API has to support usages where the tasks are homogenous or 
heterogenous, the JEP has examples of both. The type parameter is useful 
for the former, less so for the later where you have to pick a LUB or 
Object. ShutdownOnSuccess uses the type parameter as it's for racing 
subtasks that produce a result of same type. ShutdownOnFailure doesn't 
expose an API for the result so it doesn't use the type, is that your point?

-Alan