alloy-runtime-sdk 0.2.97__tar.gz → 0.2.99__tar.gz

{alloy_runtime_sdk-0.2.97 → alloy_runtime_sdk-0.2.99}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alloy-runtime-sdk
-Version: 0.2.97
+Version: 0.2.99
 Summary: Python SDK for Alloy Runtime - client, pipeline, and logging
 Keywords: alloy,api,client,python,sdk
 Classifier: Development Status :: 3 - Alpha
@@ -13,7 +13,7 @@ Description-Content-Type: text/markdown
 Requires-Dist: httpx==0.28.1
 Requires-Dist: pydantic==2.12.5
 Requires-Dist: structlog==25.5.0
-Requires-Dist: alloy-runtime-types==0.2.97
+Requires-Dist: alloy-runtime-types==0.2.99
 Requires-Dist: uuid-extension==0.2.0
 Requires-Dist: psycopg[binary]==3.3.3
 

{alloy_runtime_sdk-0.2.97 → alloy_runtime_sdk-0.2.99}/alloy_runtime_sdk/api_client/client.py

@@ -2180,12 +2180,16 @@ class ApiClient:
         timeout: float | None = None,
         poll_interval: float = 2.0,
         include_tool_call_details: bool = False,
+        max_poll_request_errors: int = 3,
     ) -> GenerateTextResponse:
         """Submit async generation, poll until terminal, and return final JSON result.
 
         By default this method has no client-side generation wait deadline. The
         server owns generation lifecycle and terminal status. Pass ``timeout``
         only when the caller explicitly wants the local wait to stop early.
+        ``max_poll_request_errors`` controls how many consecutive transient
+        status polling transport failures are allowed before the local wait
+        fails while leaving the server-side generation running.
         """
         submit = await self.generate_text_async(
             request,
@@ -2199,6 +2203,7 @@ class ApiClient:
                 timeout=timeout,
                 poll_interval=poll_interval,
                 include_tool_call_details=include_tool_call_details,
+                max_poll_request_errors=max_poll_request_errors,
             )
         except asyncio.CancelledError as exc:
             raise GenerationWaitCancelledError(execution_id) from exc
@@ -2479,6 +2484,7 @@ class ApiClient:
         timeout: float | None = None,
         poll_interval: float = 2.0,
         include_tool_call_details: bool = False,
+        max_poll_request_errors: int = 3,
     ) -> AsyncGenerationStatusResponse:
         """Wait for async generation to reach a terminal state.
 
@@ -2492,6 +2498,9 @@ class ApiClient:
                 which waits until the server reports a terminal generation state.
             poll_interval: Time between polls in seconds (default: 2.0)
             include_tool_call_details: If True, include tool call details in response
+            max_poll_request_errors: Consecutive transient polling request failures
+                to tolerate before raising. The submitted server-side generation is
+                left running when this limit is exceeded.
 
         Returns:
             Final status response when generation completes successfully
@@ -2522,15 +2531,42 @@ class ApiClient:
         start_time = time.monotonic()
         last_status: str = "pending"
         execution_id_str = str(execution_id)
+        consecutive_poll_request_errors = 0
+        if max_poll_request_errors < 1:
+            raise ValueError("max_poll_request_errors must be at least 1")
 
         while True:
             if timeout is not None and time.monotonic() - start_time >= timeout:
                 break
 
-            status_response = await self.get_generation(
-                execution_id,
-                include_tool_call_details=include_tool_call_details,
-            )
+            try:
+                status_response = await self.get_generation(
+                    execution_id,
+                    include_tool_call_details=include_tool_call_details,
+                )
+            except RequestError as exc:
+                consecutive_poll_request_errors += 1
+                if consecutive_poll_request_errors >= max_poll_request_errors:
+                    raise RequestError(
+                        f"Generation {execution_id_str} status polling failed after "
+                        f"{consecutive_poll_request_errors} consecutive request errors. "
+                        "The server-side generation may still be running; poll the "
+                        "execution ID directly before retrying the generation. "
+                        f"Last error: {exc}"
+                    ) from exc
+
+                sleep_seconds = poll_interval if poll_interval > 0 else 0.0
+                if timeout is None:
+                    await asyncio.sleep(sleep_seconds)
+                    continue
+
+                remaining = timeout - (time.monotonic() - start_time)
+                if remaining <= 0:
+                    break
+                await asyncio.sleep(min(sleep_seconds, remaining))
+                continue
+
+            consecutive_poll_request_errors = 0
             last_status = status_response.status
 
             if status_response.is_terminal:

{alloy_runtime_sdk-0.2.97 → alloy_runtime_sdk-0.2.99}/alloy_runtime_sdk/pipeline/local_runner.py

@@ -1,6 +1,7 @@
 import argparse
 import asyncio
 import json
+import shlex
 import sys
 import traceback
 from dataclasses import dataclass
@@ -218,9 +219,64 @@ def _build_parser() -> argparse.ArgumentParser:
     return parser
 
 
+def _emit_local_pipeline_resume_instructions(
+    *,
+    module_path: str,
+    pipeline_name: str | None,
+    config: str | None,
+    server_url: str,
+    pipeline_execution_id: UUID,
+    pipeline_definition_id: UUID | None,
+) -> None:
+    command_parts = [
+        "python -m alloy_runtime_sdk.pipeline.local_runner",
+        "--file",
+        module_path,
+    ]
+    if pipeline_name is not None:
+        command_parts.extend(["--function-name", pipeline_name])
+    if config is not None:
+        command_parts.extend(["--config", config])
+    command_parts.extend(
+        [
+            "--server-url",
+            server_url,
+            "--api-key",
+            "<same api key>",
+        ]
+    )
+    if pipeline_definition_id is not None:
+        command_parts.extend(["--pipeline-definition-id", str(pipeline_definition_id)])
+    command_parts.extend(["--pipeline-execution-id", str(pipeline_execution_id)])
+
+    print(
+        "\n".join(
+            [
+                "Alloy local pipeline resume instructions:",
+                f"  pipeline_execution_id: {pipeline_execution_id}",
+                "  To resume this same stop-and-go pipeline run, rerun with this exact "
+                "pipeline execution id:",
+                f"  {shlex.join(command_parts)}",
+                "  Idempotency depends on reusing this pipeline_execution_id plus stable "
+                "ctx.generate_text(step_id=...) values.",
+            ]
+        ),
+        file=sys.stderr,
+    )
+
+
 def main(argv: list[str] | None = None) -> None:
     parser = _build_parser()
     args = parser.parse_args(argv)
+    pipeline_execution_id = args.pipeline_execution_id or uuid7()
+    _emit_local_pipeline_resume_instructions(
+        module_path=args.module_path,
+        pipeline_name=args.pipeline_name,
+        config=args.config,
+        server_url=args.server_url,
+        pipeline_execution_id=pipeline_execution_id,
+        pipeline_definition_id=args.pipeline_definition_id,
+    )
 
     try:
         result = asyncio.run(
@@ -230,7 +286,7 @@ def main(argv: list[str] | None = None) -> None:
                 config=_parse_config(args.config),
                 server_url=args.server_url,
                 api_key=args.api_key,
-                pipeline_execution_id=args.pipeline_execution_id,
+                pipeline_execution_id=pipeline_execution_id,
                 pipeline_definition_id=args.pipeline_definition_id,
             )
         )

{alloy_runtime_sdk-0.2.97 → alloy_runtime_sdk-0.2.99}/alloy_runtime_sdk.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alloy-runtime-sdk
-Version: 0.2.97
+Version: 0.2.99
 Summary: Python SDK for Alloy Runtime - client, pipeline, and logging
 Keywords: alloy,api,client,python,sdk
 Classifier: Development Status :: 3 - Alpha
@@ -13,7 +13,7 @@ Description-Content-Type: text/markdown
 Requires-Dist: httpx==0.28.1
 Requires-Dist: pydantic==2.12.5
 Requires-Dist: structlog==25.5.0
-Requires-Dist: alloy-runtime-types==0.2.97
+Requires-Dist: alloy-runtime-types==0.2.99
 Requires-Dist: uuid-extension==0.2.0
 Requires-Dist: psycopg[binary]==3.3.3
 

{alloy_runtime_sdk-0.2.97 → alloy_runtime_sdk-0.2.99}/alloy_runtime_sdk.egg-info/requires.txt

@@ -1,6 +1,6 @@
 httpx==0.28.1
 pydantic==2.12.5
 structlog==25.5.0
-alloy-runtime-types==0.2.97
+alloy-runtime-types==0.2.99
 uuid-extension==0.2.0
 psycopg[binary]==3.3.3

{alloy_runtime_sdk-0.2.97 → alloy_runtime_sdk-0.2.99}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "alloy-runtime-sdk"
-version = "0.2.97"
+version = "0.2.99"
 description = "Python SDK for Alloy Runtime - client, pipeline, and logging"
 readme = "README.md"
 requires-python = ">=3.13"
@@ -20,7 +20,7 @@ dependencies = [
   "httpx==0.28.1",
   "pydantic==2.12.5",
   "structlog==25.5.0",
-  "alloy-runtime-types==0.2.97",
+  "alloy-runtime-types==0.2.99",
   "uuid-extension==0.2.0",
   "psycopg[binary]==3.3.3",
 ]