release: 0.29.0

.release-please-manifest.json

@@ -1,3 +1,3 @@
 {
-  ".": "0.28.0"
+  ".": "0.29.0"
 }

CHANGELOG.md

@@ -1,5 +1,15 @@
 # Changelog
 
+## [0.29.0](https://github.com/kernel/hypeman-python/compare/v0.28.0...v0.29.0) (2026-10-21)
+
+
+### Features
+
+* Add Windows hypervisor primitives ([5c037b9](https://github.com/kernel/hypeman-python/commit/5c037b9f9c5926025ba9e5852bea964b4e1ace7c))
+* Integrate vendor VFIO vGPUs into the instance lifecycle ([24934bb](https://github.com/kernel/hypeman-python/commit/24934bb5e6f4a2ab636e0d893d056276ce2fd3dd))
+* Place vGPU profiles the driver hides from creatable lists ([3578ec1](https://github.com/kernel/hypeman-python/commit/3578ec14db5cf79817e7e8606c9e260865deafd3))
+* Update Hypeman build toolchains to Go 1.26.10 ([6a79c69](https://github.com/kernel/hypeman-python/commit/6a79c6927c90a1b35c971dcb749d1cddc0ecf8e3))
+
 ## [0.28.0](https://github.com/kernel/hypeman-python/compare/v0.27.0...v0.28.0) (2026-09-01)
 
 

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "hypeman"
-version = "0.28.0"
+version = "0.29.0"
 description = "The official Python library for the hypeman API"
 dynamic = ["readme"]
 license = "Apache-2.0"

src/hypeman/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "hypeman"
-__version__ = "0.28.0"  # x-release-please-version
+__version__ = "0.29.0"  # x-release-please-version

src/hypeman/resources/instances/instances.py

@@ -200,16 +200,19 @@ def create(
 
           network: Network configuration for the instance
 
-          overlay_size: Writable overlay disk size (human-readable format like "10GB", "50G")
+          overlay_size: Writable overlay disk size (human-readable format like "10GB", "50G"). Omit this
+              field for macOS guests, which clone the image's complete boot disk instead.
 
-          platform: Target platform as os/arch[/variant] (e.g. "linux/amd64"), matching Docker
-              --platform. Omit for the host platform. Not a fixed enum: the os/arch[/variant]
-              grammar is validated server-side and invalid values return 400 invalid_platform.
-              Only os "linux" with arch amd64 or arm64 is accepted today.
+          platform: Target platform as os/arch[/variant]. Linux container guests support amd64 and
+              arm64. Experimental darwin/arm64 guests require an offline-imported macOS disk
+              image and vz on Apple silicon. Omit to use the cached native image or host
+              platform.
 
           restart_policy: Whole-instance restart supervision policy.
 
-          size: Base memory size (human-readable format like "1GB", "512MB", "2G")
+          size: Base memory size (human-readable format like "1GB", "512MB", "2G"). macOS guests
+              default to their image's memory; explicit sizes must be at least 4 GiB in whole
+              MiB and within VZ host limits.
 
           skip_guest_agent: Skip guest-agent installation during boot. When true, the exec and stat APIs
               will not work for this instance. The instance will still run, but remote command
@@ -230,7 +233,8 @@ def create(
               omit both expiration fields to disable automatic expiration. Mutually exclusive
               with expires_at.
 
-          vcpus: Number of virtual CPUs
+          vcpus: Number of virtual CPUs. macOS guests default to their image's vCPUs; explicit
+              counts must be at least 2 and within VZ host limits.
 
           volumes: Volumes to attach to the instance at creation time
 
@@ -536,7 +540,7 @@ def logs(
         id: str,
         *,
         follow: bool | Omit = omit,
-        source: Literal["app", "vmm", "hypeman"] | Omit = omit,
+        source: Literal["app", "vmm", "hypeman", "swtpm"] | Omit = omit,
         tail: int | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
@@ -1022,16 +1026,19 @@ async def create(
 
           network: Network configuration for the instance
 
-          overlay_size: Writable overlay disk size (human-readable format like "10GB", "50G")
+          overlay_size: Writable overlay disk size (human-readable format like "10GB", "50G"). Omit this
+              field for macOS guests, which clone the image's complete boot disk instead.
 
-          platform: Target platform as os/arch[/variant] (e.g. "linux/amd64"), matching Docker
-              --platform. Omit for the host platform. Not a fixed enum: the os/arch[/variant]
-              grammar is validated server-side and invalid values return 400 invalid_platform.
-              Only os "linux" with arch amd64 or arm64 is accepted today.
+          platform: Target platform as os/arch[/variant]. Linux container guests support amd64 and
+              arm64. Experimental darwin/arm64 guests require an offline-imported macOS disk
+              image and vz on Apple silicon. Omit to use the cached native image or host
+              platform.
 
           restart_policy: Whole-instance restart supervision policy.
 
-          size: Base memory size (human-readable format like "1GB", "512MB", "2G")
+          size: Base memory size (human-readable format like "1GB", "512MB", "2G"). macOS guests
+              default to their image's memory; explicit sizes must be at least 4 GiB in whole
+              MiB and within VZ host limits.
 
           skip_guest_agent: Skip guest-agent installation during boot. When true, the exec and stat APIs
               will not work for this instance. The instance will still run, but remote command
@@ -1052,7 +1059,8 @@ async def create(
               omit both expiration fields to disable automatic expiration. Mutually exclusive
               with expires_at.
 
-          vcpus: Number of virtual CPUs
+          vcpus: Number of virtual CPUs. macOS guests default to their image's vCPUs; explicit
+              counts must be at least 2 and within VZ host limits.
 
           volumes: Volumes to attach to the instance at creation time
 
@@ -1358,7 +1366,7 @@ async def logs(
         id: str,
         *,
         follow: bool | Omit = omit,
-        source: Literal["app", "vmm", "hypeman"] | Omit = omit,
+        source: Literal["app", "vmm", "hypeman", "swtpm"] | Omit = omit,
         tail: int | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.

src/hypeman/types/gpu_profile.py

@@ -9,7 +9,11 @@ class GPUProfile(BaseModel):
     """Available vGPU profile"""
 
     available: int
-    """Number of instances that can be created with this profile"""
+    """
+    Number of additional instances of this profile the host can take if only this
+    profile is added. Counts for different profiles are not additive: an instance of
+    one profile can reduce availability of others on the same GPU.
+    """
 
     framebuffer_mb: int
     """Frame buffer size in MB"""

src/hypeman/types/gpu_resource_status.py

@@ -13,17 +13,40 @@
 class GPUResourceStatus(BaseModel):
     """GPU resource status. Null if no GPUs available."""
 
+    allocatable_slots: int
+    """
+    Free slots eligible for placement, matching admission control (excludes
+    quarantined VFs; 0 while VF health state is unavailable)
+    """
+
     mode: Literal["vgpu", "passthrough"]
     """GPU mode (vgpu for SR-IOV/mdev, passthrough for whole GPU)"""
 
+    quarantined_slots: int
+    """VFs quarantined after guest driver init failures (vGPU mode only).
+
+    May overlap used_slots until the affected instance releases its VF.
+    """
+
     total_slots: int
     """Total slots (VFs for vGPU, physical GPUs for passthrough)"""
 
     used_slots: int
-    """Slots currently in use"""
+    """Slots currently in use.
+
+    Includes quarantined VFs that are still assigned, so this can overlap
+    quarantined_slots.
+    """
 
     devices: Optional[List[PassthroughDevice]] = None
     """Physical GPUs (only in passthrough mode)"""
 
+    placement_disabled_reason: Optional[str] = None
+    """
+    Present when allocatable_slots is 0 because the VF health state could not be
+    read or written rather than because the host is full. vGPU placement is refused
+    until the state file is repaired or the next write succeeds.
+    """
+
     profiles: Optional[List[GPUProfile]] = None
     """Available vGPU profiles (only in vGPU mode)"""

src/hypeman/types/instance.py

@@ -19,8 +19,11 @@
 class GPU(BaseModel):
     """GPU information attached to the instance"""
 
+    device_path: Optional[str] = None
+    """sysfs path of the assigned vGPU device"""
+
     mdev_uuid: Optional[str] = None
-    """mdev device UUID"""
+    """mdev device UUID (mdev hosts only)"""
 
     profile: Optional[str] = None
     """vGPU profile name"""

src/hypeman/types/instance_create_params.py

@@ -108,22 +108,29 @@ class InstanceCreateParams(TypedDict, total=False):
     """Network configuration for the instance"""
 
     overlay_size: str
-    """Writable overlay disk size (human-readable format like "10GB", "50G")"""
+    """Writable overlay disk size (human-readable format like "10GB", "50G").
+
+    Omit this field for macOS guests, which clone the image's complete boot disk
+    instead.
+    """
 
     platform: str
-    """Target platform as os/arch[/variant] (e.g.
+    """Target platform as os/arch[/variant].
 
-    "linux/amd64"), matching Docker --platform. Omit for the host platform. Not a
-    fixed enum: the os/arch[/variant] grammar is validated server-side and invalid
-    values return 400 invalid_platform. Only os "linux" with arch amd64 or arm64 is
-    accepted today.
+    Linux container guests support amd64 and arm64. Experimental darwin/arm64 guests
+    require an offline-imported macOS disk image and vz on Apple silicon. Omit to
+    use the cached native image or host platform.
     """
 
     restart_policy: RestartPolicyParam
     """Whole-instance restart supervision policy."""
 
     size: str
-    """Base memory size (human-readable format like "1GB", "512MB", "2G")"""
+    """Base memory size (human-readable format like "1GB", "512MB", "2G").
+
+    macOS guests default to their image's memory; explicit sizes must be at least 4
+    GiB in whole MiB and within VZ host limits.
+    """
 
     skip_guest_agent: bool
     """
@@ -158,7 +165,11 @@ class InstanceCreateParams(TypedDict, total=False):
     """
 
     vcpus: int
-    """Number of virtual CPUs"""
+    """Number of virtual CPUs.
+
+    macOS guests default to their image's vCPUs; explicit counts must be at least 2
+    and within VZ host limits.
+    """
 
     volumes: Iterable[VolumeMountParam]
     """Volumes to attach to the instance at creation time"""

src/hypeman/types/instance_logs_params.py

@@ -11,7 +11,7 @@ class InstanceLogsParams(TypedDict, total=False):
     follow: bool
     """Continue streaming new lines after initial output"""
 
-    source: Literal["app", "vmm", "hypeman"]
+    source: Literal["app", "vmm", "hypeman", "swtpm"]
     """Log source to stream:
 
     - app: Guest application logs (serial console output)

src/hypeman/types/instance_stats.py

@@ -32,7 +32,10 @@ class InstanceStats(BaseModel):
     """Virtual Memory Size - total virtual memory allocated in bytes"""
 
     network_rx_bytes: int
-    """Total network bytes received by the VM (from TAP interface)"""
+    """Total network bytes received by the VM (from TAP interface).
+
+    Always 0 for macOS guests, which use NAT with no per-instance interface.
+    """
 
     network_tx_bytes: int
     """Total network bytes transmitted by the VM (from TAP interface)"""