modal 1.4.4.dev21__tar.gz → 1.5.0__tar.gz

{modal-1.4.4.dev21 → modal-1.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: modal
-Version: 1.4.4.dev21
+Version: 1.5.0
 Summary: Python client library for Modal
 Author-email: Modal Labs <support@modal.com>
 License-Expression: Apache-2.0

{modal-1.4.4.dev21 → modal-1.5.0}/modal/_environments.py

@@ -104,7 +104,7 @@ class _EnvironmentManager:
 MemberRole = Literal["viewer", "contributor"]
 
 
-def _role_to_proto(role: str) -> api_pb2.EnvironmentRole.ValueType:
+def _role_to_proto(role: str) -> "api_pb2.EnvironmentRole.ValueType":
     match role:
         case "viewer":
             return api_pb2.ENVIRONMENT_ROLE_VIEWER

{modal-1.4.4.dev21 → modal-1.5.0}/modal/_server.py

@@ -34,9 +34,9 @@ def validate_http_server_config(
     if not is_server and exit_grace_period is not None and exit_grace_period > 25:
         raise InvalidError("The `exit_grace_period` argument must not exceed 25 seconds.")
 
-    if not proxy_regions:
+    if not proxy_regions or not proxy_regions[0]:
         if is_server:
-            raise InvalidError("The `routing_regions` argument must be non-empty.")
+            raise InvalidError("The `routing_region` argument must be passed.")
         raise InvalidError("The `proxy_regions` argument must be non-empty.")
 
 
@@ -49,7 +49,7 @@ class _Server:
     Instead, use the [`@app._experimental_server()`](https://modal.com/docs/reference/modal.App#server) decorator.
 
     ```python notest
-    @app._experimental_server(port=8000, routing_regions=["us-east", "us-west"])
+    @app._experimental_server(port=8000, routing_region="us-east")
     class MyServer:
         @modal.enter()
         def start_server(self):

{modal-1.4.4.dev21 → modal-1.5.0}/modal/app.py

@@ -1289,7 +1289,7 @@ class _App:
         port: int = 8000,  # Port the HTTP server listens on
         startup_timeout: int = 30,  # Maximum startup time in seconds
         exit_grace_period: int = 0,  # Grace period for in-flight requests on shutdown
-        routing_regions: list[str] = ["us-east"],  # Required: Regions to deploy proxy endpoints
+        routing_region: str = "us-east",  # Region to route Server requests through
         h2_enabled: bool = False,  # Enable HTTP/2
         target_concurrency: int | None = None,  # Target concurrency for the server; 0 disables autoscaling
         cloud: str | None = None,  # Cloud provider (aws, gcp, oci, auto)
@@ -1331,7 +1331,7 @@ class _App:
             port: Port the HTTP server listens on.
             startup_timeout: Maximum startup time in seconds.
             exit_grace_period: Grace period for in-flight requests on shutdown.
-            routing_regions: Regions to deploy proxy endpoints.
+            routing_region: Region to route Server requests through.
             h2_enabled: Enable HTTP/2.
             target_concurrency: Target concurrency for the server; 0 disables autoscaling.
             cloud: Cloud provider (aws, gcp, oci, auto).
@@ -1344,7 +1344,7 @@ class _App:
 
         Examples:
             ```python
-            @app._experimental_server(port=8000, routing_regions=["us-east"])
+            @app._experimental_server(port=8000, routing_region="us-east")
             class MyServer:
                 @modal.enter()
                 def start(self):
@@ -1361,7 +1361,7 @@ class _App:
         # Validate HTTP server config
         validate_http_server_config(
             port=port,
-            proxy_regions=routing_regions,
+            proxy_regions=[routing_region],
             startup_timeout=startup_timeout,
             exit_grace_period=exit_grace_period,
             is_server=True,
@@ -1373,7 +1373,7 @@ class _App:
 
         http_config = api_pb2.HTTPConfig(
             port=port,
-            proxy_regions=routing_regions,
+            proxy_regions=[routing_region],
             startup_timeout=startup_timeout,
             exit_grace_period=exit_grace_period,
             h2_enabled=h2_enabled,

{modal-1.4.4.dev21 → modal-1.5.0}/modal/app.pyi

@@ -752,7 +752,7 @@ class _App:
         port: int = 8000,
         startup_timeout: int = 30,
         exit_grace_period: int = 0,
-        routing_regions: list[str] = ["us-east"],
+        routing_region: str = "us-east",
         h2_enabled: bool = False,
         target_concurrency: typing.Optional[int] = None,
         cloud: typing.Optional[str] = None,
@@ -794,7 +794,7 @@ class _App:
             port: Port the HTTP server listens on.
             startup_timeout: Maximum startup time in seconds.
             exit_grace_period: Grace period for in-flight requests on shutdown.
-            routing_regions: Regions to deploy proxy endpoints.
+            routing_region: Region to route Server requests through.
             h2_enabled: Enable HTTP/2.
             target_concurrency: Target concurrency for the server; 0 disables autoscaling.
             cloud: Cloud provider (aws, gcp, oci, auto).
@@ -807,7 +807,7 @@ class _App:
 
         Examples:
             ```python
-            @app._experimental_server(port=8000, routing_regions=["us-east"])
+            @app._experimental_server(port=8000, routing_region="us-east")
             class MyServer:
                 @modal.enter()
                 def start(self):
@@ -1750,7 +1750,7 @@ class App:
         port: int = 8000,
         startup_timeout: int = 30,
         exit_grace_period: int = 0,
-        routing_regions: list[str] = ["us-east"],
+        routing_region: str = "us-east",
         h2_enabled: bool = False,
         target_concurrency: typing.Optional[int] = None,
         cloud: typing.Optional[str] = None,
@@ -1790,7 +1790,7 @@ class App:
             port: Port the HTTP server listens on.
             startup_timeout: Maximum startup time in seconds.
             exit_grace_period: Grace period for in-flight requests on shutdown.
-            routing_regions: Regions to deploy proxy endpoints.
+            routing_region: Region to route Server requests through.
             h2_enabled: Enable HTTP/2.
             target_concurrency: Target concurrency for the server; 0 disables autoscaling.
             cloud: Cloud provider (aws, gcp, oci, auto).
@@ -1803,7 +1803,7 @@ class App:
 
         Examples:
             ```python
-            @app._experimental_server(port=8000, routing_regions=["us-east"])
+            @app._experimental_server(port=8000, routing_region="us-east")
             class MyServer:
                 @modal.enter()
                 def start(self):

{modal-1.4.4.dev21 → modal-1.5.0}/modal/cli/image.py

@@ -19,12 +19,12 @@ from modal_proto import api_pb2
 
 from ._help import ModalGroup
 
-IMAGE_REGISTRY_LIST_PAGE_SIZE = 100
-IMAGE_REGISTRY_LIST_PAGE_DELAY_SECONDS = 1.0
+IMAGE_NAMES_LIST_PAGE_SIZE = 100
+IMAGE_NAMES_LIST_PAGE_DELAY_SECONDS = 1.0
 
 image_cli = ModalGroup(name="image", help="Manage Images.")
-image_registry_cli = ModalGroup(name="registry", help="Manage the Modal Image name/tag registry.")
-image_cli.add_command(image_registry_cli)
+image_names_cli = ModalGroup(name="names", help="Manage Modal Image names.")
+image_cli.add_command(image_names_cli)
 
 
 def _print_result_summary(count: int) -> None:
@@ -60,7 +60,7 @@ async def _iter_tag_pages(
             api_pb2.ImageListTagsRequest(
                 environment_name=env,
                 tag_prefix=prefix,
-                max_objects=IMAGE_REGISTRY_LIST_PAGE_SIZE,
+                max_objects=IMAGE_NAMES_LIST_PAGE_SIZE,
                 page_token=page_token,
             )
         )
@@ -70,10 +70,10 @@ async def _iter_tag_pages(
         page_token = response.next_page_token
         if not page_token:
             return
-        await asyncio.sleep(IMAGE_REGISTRY_LIST_PAGE_DELAY_SECONDS)
+        await asyncio.sleep(IMAGE_NAMES_LIST_PAGE_DELAY_SECONDS)
 
 
-@image_registry_cli.command("list", help="List named Images.")
+@image_names_cli.command("list", help="List named Images.")
 @env_option
 @click.option("--prefix", default="", help="Only include named image tags that start with this prefix.")
 @click.option("--json", is_flag=True, default=False)

{modal-1.4.4.dev21 → modal-1.5.0}/modal/cli/utils.py

@@ -1,6 +1,7 @@
 # Copyright Modal Labs 2022
 import asyncio
 import io
+import re
 import sys
 from collections.abc import Sequence
 from contextlib import nullcontext
@@ -126,6 +127,11 @@ def _plain(text: Text | str) -> str:
     return text.plain if isinstance(text, Text) else text
 
 
+def _col_name_to_json_key(name: str) -> str:
+    """Convert a display column name like "App ID" to a snake_case JSON key like "app_id"."""
+    return re.sub(r"[^a-zA-Z0-9]+", "_", name).lower().strip("_")
+
+
 def is_tty() -> bool:
     return OutputManager.get().is_terminal
 
@@ -145,7 +151,8 @@ def display_table(
 
     output = OutputManager.get()
     if json:
-        json_data = [{col_to_str(col): _plain(row[i]) for i, col in enumerate(columns)} for row in rows]
+        json_keys = [_col_name_to_json_key(col_to_str(col)) for col in columns]
+        json_data = [{json_keys[i]: _plain(row[i]) for i in range(len(columns))} for row in rows]
         output.print_json(dumps(json_data))
     elif csv:
         csv_buffer = io.StringIO()

{modal-1.4.4.dev21 → modal-1.5.0}/modal/client.pyi

@@ -31,11 +31,7 @@ class _Client:
     client_type: int
 
     def __init__(
-        self,
-        server_url: str,
-        client_type: int,
-        credentials: typing.Optional[tuple[str, str]],
-        version: str = "1.4.4.dev21",
+        self, server_url: str, client_type: int, credentials: typing.Optional[tuple[str, str]], version: str = "1.5.0"
     ):
         """mdmd:hidden
         The Modal client object is not intended to be instantiated directly by users.
@@ -201,11 +197,7 @@ class Client:
     client_type: int
 
     def __init__(
-        self,
-        server_url: str,
-        client_type: int,
-        credentials: typing.Optional[tuple[str, str]],
-        version: str = "1.4.4.dev21",
+        self, server_url: str, client_type: int, credentials: typing.Optional[tuple[str, str]], version: str = "1.5.0"
     ):
         """mdmd:hidden
         The Modal client object is not intended to be instantiated directly by users.

{modal-1.4.4.dev21 → modal-1.5.0}/modal/sandbox.py

@@ -3,12 +3,13 @@ import asyncio
 import builtins
 import enum
 import json
+import logging
 import os
 import time
 import typing
 import uuid
 import weakref
-from collections.abc import AsyncGenerator, AsyncIterator, Collection, Sequence
+from collections.abc import AsyncGenerator, AsyncIterator, Awaitable, Collection, Sequence
 from dataclasses import dataclass
 from pathlib import PurePosixPath
 from typing import TYPE_CHECKING, Any, Literal, Union, overload
@@ -71,6 +72,42 @@ from .stream_type import StreamType
 _default_image: _Image = _Image.debian_slim()
 
 
+async def _gather_load_with_timings(
+    load_coros: Sequence[Awaitable[Any]],
+) -> list[tuple[str, float]]:
+    """Await all loader coroutines concurrently and return [(object_id, elapsed_seconds)] per load."""
+    timings: list[tuple[str, float]] = []
+
+    async def timed(coro: Awaitable[Any]) -> None:
+        start = time.monotonic()
+        obj = await coro
+        timings.append((obj.object_id, time.monotonic() - start))
+
+    await asyncio.gather(*(timed(c) for c in load_coros))
+    return timings
+
+
+def _format_sandbox_create_timing_log(
+    sandbox_id: str,
+    total_seconds: float,
+    rpc_seconds: float,
+    dep_timings: Sequence[tuple[str, float]],
+) -> str:
+    """Format the Sandbox create debug log line, listing the slowest deps first."""
+    if dep_timings:
+        deps_sorted = sorted(dep_timings, key=lambda t: t[1], reverse=True)
+        shown = deps_sorted[:10]
+        dep_summary = ", ".join(f"{label}: {elapsed:.2f}s" for label, elapsed in shown)
+        if len(deps_sorted) > 10:
+            dep_summary += f", +{len(deps_sorted) - 10} more"
+    else:
+        dep_summary = "none"
+    return (
+        f"Sandbox {sandbox_id} created in {total_seconds:.2f}s "
+        f"(create rpc: {rpc_seconds:.2f}s; dependencies: {dep_summary})"
+    )
+
+
 # The maximum number of bytes that can be passed to an exec on Linux.
 # Though this is technically a 'server side' limit, it is unlikely to change.
 # getconf ARG_MAX will show this value on a host.
@@ -349,6 +386,7 @@ class _Sandbox(_Object, type_prefix="sb"):
             pty_info = _Sandbox._default_pty_info()
 
         async def _load(self: _Sandbox, resolver: Resolver, load_context: LoadContext, _existing_object_id: str | None):
+            load_start = time.monotonic()
             # An already-hydrated image (e.g. one returned by
             # `Sandbox.snapshot_directory`) is skipped — there's nothing to load.
             dep_tasks: list = []
@@ -365,8 +403,7 @@ class _Sandbox(_Object, type_prefix="sb"):
                     dep_tasks.append(resolver.load(cloud_bucket_mount.secret, load_context))
             if proxy:
                 dep_tasks.append(resolver.load(proxy, load_context))
-            if dep_tasks:
-                await asyncio.gather(*dep_tasks)
+            dep_timings = await _gather_load_with_timings(dep_tasks) if dep_tasks else []
 
             # Validate that the same volume (by object_id) isn't mounted at multiple paths
             validate_volumes_by_object_id(validated_volumes)
@@ -444,10 +481,16 @@ class _Sandbox(_Object, type_prefix="sb"):
             create_req = api_pb2.SandboxCreateRequest(
                 app_id=load_context.app_id, definition=definition, tags=tag_protos
             )
+            rpc_start = time.monotonic()
             create_resp = await load_context.client.stub.SandboxCreate(create_req)
+            rpc_elapsed = time.monotonic() - rpc_start
             sandbox_id = create_resp.sandbox_id
             self._hydrate(sandbox_id, load_context.client, None)
 
+            if logger.isEnabledFor(logging.DEBUG):
+                total_elapsed = time.monotonic() - load_start
+                logger.debug(_format_sandbox_create_timing_log(sandbox_id, total_elapsed, rpc_elapsed, dep_timings))
+
         return _Sandbox._from_loader(_load, "Sandbox()", load_context_overrides=LoadContext.empty())
 
     @staticmethod
@@ -864,6 +907,7 @@ class _Sandbox(_Object, type_prefix="sb"):
             )
 
         async def _load(self: _Sandbox, resolver: Resolver, load_context: LoadContext, _existing_object_id: str | None):
+            load_start = time.monotonic()
             dep_tasks: list = []
             if not image._is_hydrated:
                 dep_tasks.append(resolver.load(image, load_context))
@@ -874,8 +918,7 @@ class _Sandbox(_Object, type_prefix="sb"):
             for _, cloud_bucket_mount in cloud_bucket_mounts:
                 if cloud_bucket_mount.secret:
                     dep_tasks.append(resolver.load(cloud_bucket_mount.secret, load_context))
-            if dep_tasks:
-                await asyncio.gather(*dep_tasks)
+            dep_timings = await _gather_load_with_timings(dep_tasks) if dep_tasks else []
 
             validate_volumes_by_object_id(validated_volumes)
 
@@ -909,9 +952,11 @@ class _Sandbox(_Object, type_prefix="sb"):
             create_req = api_pb2.SandboxCreateV2Request(app_id=load_context.app_id, definition=definition)
             assert load_context.client._auth_token_manager
             auth_token = await load_context.client._auth_token_manager.get_token()
+            rpc_start = time.monotonic()
             create_resp = await load_context.client.stub.SandboxCreateV2(
                 create_req, metadata=[("x-modal-auth-token", auth_token)]
             )
+            rpc_elapsed = time.monotonic() - rpc_start
             sandbox_id = create_resp.sandbox_id
             self._hydrate(sandbox_id, load_context.client, None)
             self._is_v2 = True
@@ -922,6 +967,10 @@ class _Sandbox(_Object, type_prefix="sb"):
                 for t in create_resp.tunnels
             }
 
+            if logger.isEnabledFor(logging.DEBUG):
+                total_elapsed = time.monotonic() - load_start
+                logger.debug(_format_sandbox_create_timing_log(sandbox_id, total_elapsed, rpc_elapsed, dep_timings))
+
         obj = _Sandbox._from_loader(_load, "Sandbox()", load_context_overrides=LoadContext.empty())
 
         app_id: str | None = None
@@ -1247,8 +1296,6 @@ class _Sandbox(_Object, type_prefix="sb"):
             sandbox_session_2.filesystem.list_files("/user_project")
             ```
         """
-        self._ensure_v1("mount_image")
-
         if not isinstance(image, _Image):
             raise TypeError(f"Sandbox.mount_image(image=...) expects an Image object, got {image!r}")
 
@@ -1299,8 +1346,6 @@ class _Sandbox(_Object, type_prefix="sb"):
             path: Absolute mount point directory to unmount.
 
         """
-        self._ensure_v1("unmount_image")
-
         task_id = await self._get_task_id()
         command_router_client = await self._get_command_router_client(task_id)
 
@@ -1344,8 +1389,6 @@ class _Sandbox(_Object, type_prefix="sb"):
             sandbox_session_2.filesystem.list_files("/user_project")
             ```
         """
-        self._ensure_v1("snapshot_directory")
-
         wire_ttl_seconds = _ttl_to_wire_ttl(ttl)
 
         task_id = await self._get_task_id()

{modal-1.4.4.dev21 → modal-1.5.0}/modal/sandbox.pyi

@@ -29,6 +29,18 @@ import pathlib
 import typing
 import typing_extensions
 
+async def _gather_load_with_timings(
+    load_coros: collections.abc.Sequence[collections.abc.Awaitable[typing.Any]],
+) -> list[tuple[str, float]]:
+    """Await all loader coroutines concurrently and return [(object_id, elapsed_seconds)] per load."""
+    ...
+
+def _format_sandbox_create_timing_log(
+    sandbox_id: str, total_seconds: float, rpc_seconds: float, dep_timings: collections.abc.Sequence[tuple[str, float]]
+) -> str:
+    """Format the Sandbox create debug log line, listing the slowest deps first."""
+    ...
+
 def _ttl_to_wire_ttl(ttl: typing.Optional[int]) -> int:
     """Convert a TTL value to the wire format, validating the input."""
     ...

{modal-1.4.4.dev21 → modal-1.5.0}/modal/server.pyi

@@ -14,7 +14,7 @@ class Server:
     Instead, use the [`@app._experimental_server()`](https://modal.com/docs/reference/modal.App#server) decorator.
 
     ```python notest
-    @app._experimental_server(port=8000, routing_regions=["us-east", "us-west"])
+    @app._experimental_server(port=8000, routing_region="us-east")
     class MyServer:
         @modal.enter()
         def start_server(self):

{modal-1.4.4.dev21 → modal-1.5.0}/modal.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: modal
-Version: 1.4.4.dev21
+Version: 1.5.0
 Summary: Python client library for Modal
 Author-email: Modal Labs <support@modal.com>
 License-Expression: Apache-2.0