Add support for host features.

[PiotrSikora]

The proxy_get_host_features hostcall can be used by the SDKs to
determine which features are supported by the host environment.

[PiotrSikora]

As mentioned in the meeting, this is still missing text about hosts generating UNIMPLEMENTED stubs for any imports starting with proxy_ to avoid linking-time issues during loading phase for imports supported/expected by guests but not implemented yet by the host.

[PiotrSikora]

@mpwarres@leonm1 I believe all comments and discussion from the meetings have been addressed here (I left the comments open, so please resolve them if you're satisfied with the answers).

[leonm1]

Looks great, thanks!

[mpwarres]

LGTM, thanks!

[leonm1]

LGTM for merging.

I propose we merge this as-is and handle guest required features in a different PR, so we can build on HOST_FEATURES.md in other PRs.

[PiotrSikora]

@leonm1@mpwarres we now have:

• NOT_SUPPORTED for unsupported features - either because of missing features, policy, or possibly other reasons,
• UNIMPLEMENTED for proxy_continue_stream(Upstream) in v0.2.

Do we want to switch that UNIMPLEMENTED to more generic NOT_SUPPORTED?

[mpwarres]

Switching it to NOT_SUPPORTED seems desirable for uniformity, I guess the concern is about backwards compatibility? It would only be a change to the return code in the failure case, so I think that change is unlikely to be disruptive.

Is there a meaningful distinction between UNIMPLEMENTED and NOT_SUPPORTED, though? If it's just that NOT_SUPPORTED is more general, how about renaming value 12 from UNIMPLEMENTED to NOT_SUPPORTED (and maybe retaining an enum alias)?

[PiotrSikora]

The proxy_get_host_features hostcall can be used by the SDKs to
determine which features are supported by the host environment.

[PiotrSikora]

As mentioned in the meeting, this is still missing text about hosts generating UNIMPLEMENTED stubs for any imports starting with proxy_ to avoid linking-time issues during loading phase for imports supported/expected by guests but not implemented yet by the host.

[PiotrSikora]

@mpwarres@leonm1 I believe all comments and discussion from the meetings have been addressed here (I left the comments open, so please resolve them if you're satisfied with the answers).

[leonm1]

Looks great, thanks!

[mpwarres]

LGTM, thanks!

[leonm1]

LGTM for merging.

I propose we merge this as-is and handle guest required features in a different PR, so we can build on HOST_FEATURES.md in other PRs.

[PiotrSikora]

@leonm1@mpwarres we now have:

• NOT_SUPPORTED for unsupported features - either because of missing features, policy, or possibly other reasons,
• UNIMPLEMENTED for proxy_continue_stream(Upstream) in v0.2.

Do we want to switch that UNIMPLEMENTED to more generic NOT_SUPPORTED?

[mpwarres]

Switching it to NOT_SUPPORTED seems desirable for uniformity, I guess the concern is about backwards compatibility? It would only be a change to the return code in the failure case, so I think that change is unlikely to be disruptive.

Is there a meaningful distinction between UNIMPLEMENTED and NOT_SUPPORTED, though? If it's just that NOT_SUPPORTED is more general, how about renaming value 12 from UNIMPLEMENTED to NOT_SUPPORTED (and maybe retaining an enum alias)?

[PiotrSikora]

The proxy_get_host_features hostcall can be used by the SDKs to
determine which features are supported by the host environment.

[PiotrSikora]

As mentioned in the meeting, this is still missing text about hosts generating UNIMPLEMENTED stubs for any imports starting with proxy_ to avoid linking-time issues during loading phase for imports supported/expected by guests but not implemented yet by the host.

[PiotrSikora]

@mpwarres@leonm1 I believe all comments and discussion from the meetings have been addressed here (I left the comments open, so please resolve them if you're satisfied with the answers).

[leonm1]

Looks great, thanks!

[mpwarres]

LGTM, thanks!

[leonm1]

LGTM for merging.

I propose we merge this as-is and handle guest required features in a different PR, so we can build on HOST_FEATURES.md in other PRs.

[PiotrSikora]

@leonm1@mpwarres we now have:

• NOT_SUPPORTED for unsupported features - either because of missing features, policy, or possibly other reasons,
• UNIMPLEMENTED for proxy_continue_stream(Upstream) in v0.2.

Do we want to switch that UNIMPLEMENTED to more generic NOT_SUPPORTED?

[mpwarres]

Switching it to NOT_SUPPORTED seems desirable for uniformity, I guess the concern is about backwards compatibility? It would only be a change to the return code in the failure case, so I think that change is unlikely to be disruptive.

Is there a meaningful distinction between UNIMPLEMENTED and NOT_SUPPORTED, though? If it's just that NOT_SUPPORTED is more general, how about renaming value 12 from UNIMPLEMENTED to NOT_SUPPORTED (and maybe retaining an enum alias)?

[PiotrSikora]

The proxy_get_host_features hostcall can be used by the SDKs to
determine which features are supported by the host environment.

[PiotrSikora]

As mentioned in the meeting, this is still missing text about hosts generating UNIMPLEMENTED stubs for any imports starting with proxy_ to avoid linking-time issues during loading phase for imports supported/expected by guests but not implemented yet by the host.

[PiotrSikora]

@mpwarres@leonm1 I believe all comments and discussion from the meetings have been addressed here (I left the comments open, so please resolve them if you're satisfied with the answers).

[leonm1]

Looks great, thanks!

[mpwarres]

LGTM, thanks!

[leonm1]

LGTM for merging.

I propose we merge this as-is and handle guest required features in a different PR, so we can build on HOST_FEATURES.md in other PRs.

[PiotrSikora]

@leonm1@mpwarres we now have:

• NOT_SUPPORTED for unsupported features - either because of missing features, policy, or possibly other reasons,
• UNIMPLEMENTED for proxy_continue_stream(Upstream) in v0.2.

Do we want to switch that UNIMPLEMENTED to more generic NOT_SUPPORTED?

[mpwarres]

Switching it to NOT_SUPPORTED seems desirable for uniformity, I guess the concern is about backwards compatibility? It would only be a change to the return code in the failure case, so I think that change is unlikely to be disruptive.

Is there a meaningful distinction between UNIMPLEMENTED and NOT_SUPPORTED, though? If it's just that NOT_SUPPORTED is more general, how about renaming value 12 from UNIMPLEMENTED to NOT_SUPPORTED (and maybe retaining an enum alias)?

[PiotrSikora]

The proxy_get_host_features hostcall can be used by the SDKs to
determine which features are supported by the host environment.

[PiotrSikora]

As mentioned in the meeting, this is still missing text about hosts generating UNIMPLEMENTED stubs for any imports starting with proxy_ to avoid linking-time issues during loading phase for imports supported/expected by guests but not implemented yet by the host.

[PiotrSikora]

@mpwarres@leonm1 I believe all comments and discussion from the meetings have been addressed here (I left the comments open, so please resolve them if you're satisfied with the answers).

[leonm1]

Looks great, thanks!

[mpwarres]

LGTM, thanks!

[leonm1]

LGTM for merging.

I propose we merge this as-is and handle guest required features in a different PR, so we can build on HOST_FEATURES.md in other PRs.

[PiotrSikora]

@leonm1@mpwarres we now have:

• NOT_SUPPORTED for unsupported features - either because of missing features, policy, or possibly other reasons,
• UNIMPLEMENTED for proxy_continue_stream(Upstream) in v0.2.

Do we want to switch that UNIMPLEMENTED to more generic NOT_SUPPORTED?

[mpwarres]

Switching it to NOT_SUPPORTED seems desirable for uniformity, I guess the concern is about backwards compatibility? It would only be a change to the return code in the failure case, so I think that change is unlikely to be disruptive.

Is there a meaningful distinction between UNIMPLEMENTED and NOT_SUPPORTED, though? If it's just that NOT_SUPPORTED is more general, how about renaming value 12 from UNIMPLEMENTED to NOT_SUPPORTED (and maybe retaining an enum alias)?

[PiotrSikora]

The proxy_get_host_features hostcall can be used by the SDKs to
determine which features are supported by the host environment.

[PiotrSikora]

As mentioned in the meeting, this is still missing text about hosts generating UNIMPLEMENTED stubs for any imports starting with proxy_ to avoid linking-time issues during loading phase for imports supported/expected by guests but not implemented yet by the host.

[PiotrSikora]

@mpwarres@leonm1 I believe all comments and discussion from the meetings have been addressed here (I left the comments open, so please resolve them if you're satisfied with the answers).

[leonm1]

Looks great, thanks!

[mpwarres]

LGTM, thanks!

[leonm1]

LGTM for merging.

I propose we merge this as-is and handle guest required features in a different PR, so we can build on HOST_FEATURES.md in other PRs.

[PiotrSikora]

@leonm1@mpwarres we now have:

• NOT_SUPPORTED for unsupported features - either because of missing features, policy, or possibly other reasons,
• UNIMPLEMENTED for proxy_continue_stream(Upstream) in v0.2.

Do we want to switch that UNIMPLEMENTED to more generic NOT_SUPPORTED?

[mpwarres]

Switching it to NOT_SUPPORTED seems desirable for uniformity, I guess the concern is about backwards compatibility? It would only be a change to the return code in the failure case, so I think that change is unlikely to be disruptive.

Is there a meaningful distinction between UNIMPLEMENTED and NOT_SUPPORTED, though? If it's just that NOT_SUPPORTED is more general, how about renaming value 12 from UNIMPLEMENTED to NOT_SUPPORTED (and maybe retaining an enum alias)?

[PiotrSikora]

The proxy_get_host_features hostcall can be used by the SDKs to
determine which features are supported by the host environment.

[PiotrSikora]

As mentioned in the meeting, this is still missing text about hosts generating UNIMPLEMENTED stubs for any imports starting with proxy_ to avoid linking-time issues during loading phase for imports supported/expected by guests but not implemented yet by the host.

[PiotrSikora]

@mpwarres@leonm1 I believe all comments and discussion from the meetings have been addressed here (I left the comments open, so please resolve them if you're satisfied with the answers).

[leonm1]

Looks great, thanks!

[mpwarres]

LGTM, thanks!

[leonm1]

LGTM for merging.

I propose we merge this as-is and handle guest required features in a different PR, so we can build on HOST_FEATURES.md in other PRs.

[PiotrSikora]

@leonm1@mpwarres we now have:

• NOT_SUPPORTED for unsupported features - either because of missing features, policy, or possibly other reasons,
• UNIMPLEMENTED for proxy_continue_stream(Upstream) in v0.2.

Do we want to switch that UNIMPLEMENTED to more generic NOT_SUPPORTED?

[mpwarres]

Switching it to NOT_SUPPORTED seems desirable for uniformity, I guess the concern is about backwards compatibility? It would only be a change to the return code in the failure case, so I think that change is unlikely to be disruptive.

Is there a meaningful distinction between UNIMPLEMENTED and NOT_SUPPORTED, though? If it's just that NOT_SUPPORTED is more general, how about renaming value 12 from UNIMPLEMENTED to NOT_SUPPORTED (and maybe retaining an enum alias)?

[PiotrSikora]

The proxy_get_host_features hostcall can be used by the SDKs to
determine which features are supported by the host environment.

[PiotrSikora]

As mentioned in the meeting, this is still missing text about hosts generating UNIMPLEMENTED stubs for any imports starting with proxy_ to avoid linking-time issues during loading phase for imports supported/expected by guests but not implemented yet by the host.

[PiotrSikora]

@mpwarres@leonm1 I believe all comments and discussion from the meetings have been addressed here (I left the comments open, so please resolve them if you're satisfied with the answers).

[leonm1]

Looks great, thanks!

[mpwarres]

LGTM, thanks!

[leonm1]

LGTM for merging.

I propose we merge this as-is and handle guest required features in a different PR, so we can build on HOST_FEATURES.md in other PRs.

[PiotrSikora]

@leonm1@mpwarres we now have:

• NOT_SUPPORTED for unsupported features - either because of missing features, policy, or possibly other reasons,
• UNIMPLEMENTED for proxy_continue_stream(Upstream) in v0.2.

Do we want to switch that UNIMPLEMENTED to more generic NOT_SUPPORTED?

[mpwarres]

Switching it to NOT_SUPPORTED seems desirable for uniformity, I guess the concern is about backwards compatibility? It would only be a change to the return code in the failure case, so I think that change is unlikely to be disruptive.

Is there a meaningful distinction between UNIMPLEMENTED and NOT_SUPPORTED, though? If it's just that NOT_SUPPORTED is more general, how about renaming value 12 from UNIMPLEMENTED to NOT_SUPPORTED (and maybe retaining an enum alias)?