WIP: Custom backends for Dynamic Selection #2220
Conversation
| The `get_submission_group` is not easily implemented in a meaningful way. It must return a type that | ||
| defines a member function `wait`. But since there is no way to know how to wait on all previous submissions | ||
| for an arbitrary backend resource, this function will likely need to return a dummy type or | ||
| `get_submission_group` be undefined for the default backend. |
There was a problem hiding this comment.
If we could provide a way to have the user give us the wait type, do we have a shot? Or maybe it's better to leave this out to make the default case simple.
There was a problem hiding this comment.
Lets consider if there's a meaningful default or not. I'm not sure that there is.
There was a problem hiding this comment.
I think a meaningful default is to call wait on all resources, if they have a wait function.
| Policies contain a backend and current policies are either default constructed/initialized or | ||
| constructed/initialized with a `std::vector<Resouce>` of resources. When a vector is passed | ||
| to the constructor, the type of the resource can be deduced and used as a template argument | ||
| when constructing the backend. When a policy is default constructed, the resource type can be made | ||
| a manditory template argument and this can be used to set the resource type in the backend. |
There was a problem hiding this comment.
Should include deferred initialization here.
| when constructing the backend. When a policy is default constructed, the resource type can be made | ||
| a manditory template argument and this can be used to set the resource type in the backend. | ||
|
|
||
| Is it therefore becomes unnecessary to explicitly provide the resource type in the default backend. |
There was a problem hiding this comment.
| Is it therefore becomes unnecessary to explicitly provide the resource type in the default backend. | |
| It therefore becomes unnecessary to explicitly provide the resource type in the default backend. |
There was a problem hiding this comment.
Also, we probably want to soften the language here a bit to accomodate the default constructor and deferred initialization constructors.
Perhaps we can just add on "when providing resource objects to a constructor." or something like that.
|
|
||
| // v is empty | ||
| auto v = p.get_resources(); | ||
| ``` |
There was a problem hiding this comment.
mention deferred initialization.
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
| namespace experimental { | ||
|
|
||
| template <typename ResourceType> | ||
| struct backend_for_resource |
There was a problem hiding this comment.
I know this is my naming, but I don't like it, perhaps it should just be resource_traits , perhaps backend_t should be default_backend_t?
resource_traits<resource_t>::default_backend_t
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
Co-authored-by: Dan Hoeflinger <109972525+danhoeflinger@users.noreply.github.com>
A very rough draft for custom backends for dynamic selection.