feat: restore legacy recall, deletion, queue and workflow-form behaviors

docs/src/content/docs/Users Guide/04.Image Generation/01.introduction.mdx

@@ -49,7 +49,9 @@ If a required component is missing, **Invoke** is disabled and its tooltip names
 
 </Steps>
 
-Hover over **Invoke** to see what will be queued, for example *1 prompt × 4 iterations → 4 generations*. When Invoke is disabled, the same tooltip lists what is stopping it.
+Hover over **Invoke** to see what will be queued, for example *1 prompt × 4 iterations → 4 generations*. When Invoke is disabled, the same tooltip lists what is stopping it. To run a batch ahead of the work you already have queued, press **Ctrl+Shift+Enter** (**Cmd+Shift+Enter** on macOS) or choose **Invoke (Front of Queue)** from the command palette. When several people share a server, the default round-robin queue still alternates between users, so the batch only moves ahead of your own jobs; a server set to first-in, first-out (`session_queue_mode: FIFO`) runs it ahead of everyone's pending jobs.
+
+The **Queue** panel lists recent jobs. Cancel a pending, waiting or running job from its row, and retry a failed or canceled job from the session the server saved for it, independently of your current settings. A retried job's results go where the original was headed while the tab that sent it is still open: back to its place in the Canvas staging area, or to the Gallery board it was sent to. This also holds when you retry several failed items of one batch one after another. A retried Canvas job is staged in a tab where its project is open; if that project is open nowhere, its result does not appear. Once that tab is gone, for example after a reload, or once its Canvas was replaced with a new document, Canvas results are staged once, at the current bounding box, by a tab with the project open, and Gallery results appear in the Gallery, on **Uncategorized** unless the job itself saved them to a board. Expand a row to cancel its whole batch while other batches keep running, or to open **View JSON**, which shows the graph the server actually ran, its results and the embedded workflow. The panel menu clears failed items, clears every finished item while keeping queued and running work, or clears the whole queue. Administrators also see the **Invocation Cache** beneath the queue, next to the model cache, where cached node outputs can be cleared, disabled or re-enabled.
 
 ## Routing images to the Gallery or the Canvas
 

docs/src/content/docs/Users Guide/06.Gallery/introduction.mdx

@@ -148,9 +148,9 @@ Drag items onto a board in the board list; if the list is hidden, it opens while
 
 ### Deleting media
 
-Press **Delete** (or **Backspace**), choose **Delete Image**, **Delete Video** or **Delete Selection** from the menu, or use the delete button in the selection bar. Deleting permanently removes the items from every board and from disk, and cannot be undone. A confirmation asks first. Tick **Don't ask me again** and confirm to stop asking (cancelling leaves it on); you can turn it back on, or off, with **Confirm image deletion** in Settings under Gallery.
+Press **Delete** (or **Backspace**), choose **Delete Image**, **Delete Video** or **Delete Selection** from the menu, or use the delete button in the selection bar. Deleting permanently removes the items from every board and from disk, and cannot be undone. A confirmation asks first. Tick **Don't ask me again** and confirm to stop asking for ordinary deletions (cancelling leaves it on); you can turn it back on, or off, with **Confirm image deletion** in Settings under Gallery.
 
-Deleting does not check whether an image is still used elsewhere, for example on a canvas or as a reference image, so check before you delete media you are still working with. If deleting is interrupted, InvokeAI cleans up any leftover files the next time it starts.
+When the open project still uses something you are deleting, for example as a canvas layer, a reference image, an upscale or video input, or in a workflow, the confirmation always appears and lists where, even if you turned confirmations off. Only the open project's current work is checked: other projects, undo history and canvas snapshots are not, so check before you delete media you still need there. If deleting is interrupted, InvokeAI cleans up any leftover files the next time it starts.
 
 ## Related pages
 

docs/src/content/docs/Users Guide/08.Workflows/01.introduction.mdx

@@ -54,7 +54,7 @@ Hover over a numbered circle, or tap it on a touchscreen, for a short descriptio
 |---|---|---|
 | 1 | **Workflow panel** | The workflow's **form**: its most important settings as simple controls, like a Generate panel for this workflow. |
 | 2 | **View** / **Edit** | **View** uses the form. **Edit** lets you design the form, edit the workflow's details (name, description, author, tags, notes), and inspect the selected node. |
-| 3 | **Editor toolbar** | Tools for panning, box and lasso selection, and erasing nodes or connections; **Zoom in**, **Zoom out** and **Fit view**; **Download workflow as PNG**; and **Node opacity**. |
+| 3 | **Editor toolbar** | Tools for panning, box and lasso selection, and erasing nodes or connections; **Zoom in**, **Zoom out** and **Fit view**; **Arrange nodes**, which lays out the whole graph, or just the selected nodes in place, as one undoable step; **Download workflow as PNG**; and **Node opacity**. |
 | 4 | **Workflow name** | The active workflow. Click it to open the **Workflows** library. |
 | 5 | **Workflow actions** | **Add node** (**+**), **Save to library…**, the save status, settings, and the **⋯** menu with **Workflow details…**, **Rename workflow…**, **Import workflow JSON…**, **Export workflow JSON**, **New workflow** and more. |
 | 6 | **The graph** | The workflow's nodes and connections. Drag nodes by their title bar; scroll to zoom; drag empty space to pan. |

docs/src/content/docs/configuration/Multi-User Mode/admin-guide.mdx

@@ -68,7 +68,7 @@ INVOKEAI_SESSION_QUEUE_MODE=FIFO
 `session_queue_mode` only applies in multiuser mode. In single-user mode the queue is always FIFO regardless of this setting, since all jobs belong to the same account.
 :::
 
-Round-robin fairness is determined from when each user's jobs were last started. Retained terminal queue history (see [`max_queue_history`](#configuration-reference)) does not slow dequeue scheduling — each dequeue's cost scales with the number of users who currently have pending jobs, not with the size of the history.
+Round-robin fairness is determined from when each user's jobs were last started. Retained terminal queue history (see [`max_queue_history`](#configuration-reference)) does not slow dequeue scheduling — each dequeue's cost scales with the number of users who currently have pending jobs, not with the size of the history. Administrators can change `max_queue_history` without editing the file: open **Settings → Server → Queue history**, where a blank value keeps every finished item and `0` prunes them all at the next start.
 
 ### First Administrator Account
 

invokeai/app/api/routers/session_queue.py

@@ -29,6 +29,8 @@
     EnqueueReceiptLimitError,
     ItemIdsResult,
     PruneResult,
+    QueueItemDeliveryRequest,
+    QueueItemDeliveryResult,
     RetryItemsResult,
     SessionQueueCountsByDestination,
     SessionQueueItem,
@@ -882,6 +884,40 @@ def cancel_queue_item(
         raise HTTPException(status_code=500, detail=f"Unexpected error while canceling queue item: {e}")
 
 
+@session_queue_router.post(
+    "/{queue_id}/i/{item_id}/delivery",
+    operation_id="claim_queue_item_delivery",
+    response_model=QueueItemDeliveryResult,
+)
+def claim_queue_item_delivery(
+    current_user: CurrentUserOrDefault,
+    queue_id: str = Path(description="The queue id to perform this operation on"),
+    item_id: int = Path(description="The queue item whose results are delivered"),
+    request: QueueItemDeliveryRequest = Body(description="The delivery change"),
+) -> QueueItemDeliveryResult:
+    """Claims, confirms or releases staging a queue item's results, so that only one of its owner's browsers stages
+    them. A claim is held for a while unless confirmed, released or renewed."""
+    session_queue = ApiDependencies.invoker.services.session_queue
+    try:
+        # Waiting browsers ask often: authorize by the item's owner alone, without reading its session.
+        owner = session_queue.get_queue_item_owner(item_id)
+        if owner.queue_id != queue_id:
+            raise HTTPException(status_code=404, detail=f"Queue item with id {item_id} not found in queue {queue_id}")
+        if owner.user_id != current_user.user_id and not current_user.is_admin:
+            raise HTTPException(status_code=403, detail=f"You do not have permission to mutate queue item {item_id}")
+        if request.action == "claim":
+            return session_queue.claim_queue_item_delivery(item_id, request.claimant)
+        if request.action == "confirm":
+            return session_queue.confirm_queue_item_delivery(item_id)
+        return session_queue.release_queue_item_delivery(item_id, request.claimant)
+    except SessionQueueItemNotFoundError:
+        raise HTTPException(status_code=404, detail=f"Queue item with id {item_id} not found in queue {queue_id}")
+    except HTTPException:
+        raise
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Unexpected error while changing queue item delivery: {e}")
+
+
 @session_queue_router.get(
     "/{queue_id}/counts_by_destination",
     operation_id="counts_by_destination",

invokeai/app/api/sockets.py

@@ -920,16 +920,24 @@ async def _emit_bulk_queue_item_event(
         """
         item_ids_by_user: dict[str, list[int]] = getattr(event_data, item_ids_by_user_field)
         admin_sids = self._admin_sids()
+
+        def scoped(item_ids: list[int], user_ids: list[str]) -> dict[str, Any]:
+            update: dict[str, Any] = {
+                item_ids_field: item_ids,
+                "user_ids": user_ids,
+                item_ids_by_user_field: dict.fromkeys(user_ids, item_ids),
+            }
+            if isinstance(event_data, QueueItemsRetriedEvent):
+                # A retry's new item ids are as private as the retried ids they are paired with.
+                owned_item_ids = set(item_ids)
+                update["retried_items"] = [
+                    item for item in event_data.retried_items if item.source_item_id in owned_item_ids
+                ]
+            return update
+
         for user_id in event_data.user_ids:
             user_room = f"user:{user_id}"
-            owner_item_ids = item_ids_by_user.get(user_id, [])
-            owner_event_data = event_data.model_copy(
-                update={
-                    item_ids_field: owner_item_ids,
-                    "user_ids": [user_id],
-                    item_ids_by_user_field: {user_id: owner_item_ids},
-                }
-            )
+            owner_event_data = event_data.model_copy(update=scoped(item_ids_by_user.get(user_id, []), [user_id]))
             await self._sio.emit(
                 event=event_name,
                 data=owner_event_data.model_dump(mode="json"),
@@ -938,7 +946,7 @@ async def _emit_bulk_queue_item_event(
             )
         await self._sio.emit(event=event_name, data=event_data.model_dump(mode="json"), room="admin")
 
-        sanitized = event_data.model_copy(update={item_ids_field: [], "user_ids": [], item_ids_by_user_field: {}})
+        sanitized = event_data.model_copy(update=scoped([], []))
         await self._sio.emit(
             event=event_name,
             data=sanitized.model_dump(mode="json"),

invokeai/app/services/events/events_common.py

@@ -13,6 +13,7 @@
     QUEUE_ITEM_STATUS,
     BatchStatus,
     EnqueueBatchResult,
+    RetriedQueueItem,
     RetryItemsResult,
     SessionQueueItem,
     SessionQueueStatus,
@@ -384,6 +385,9 @@ class QueueItemsRetriedEvent(QueueEventBase):
     retried_item_ids_by_user: dict[str, list[int]] = Field(
         description="The retried root queue item IDs keyed by owner user ID."
     )
+    retried_items: list[RetriedQueueItem] = Field(
+        default_factory=list, description="Each retried root queue item with the queue item that re-runs it"
+    )
 
     @classmethod
     def build(
@@ -394,6 +398,7 @@ def build(
             retried_item_ids=retry_result.retried_item_ids,
             user_ids=user_ids,
             retried_item_ids_by_user=retried_item_ids_by_user,
+            retried_items=retry_result.retried_items,
         )
 
 

invokeai/app/services/session_queue/session_queue_base.py

@@ -20,6 +20,8 @@
     ItemIdsResult,
     NodeFieldValue,
     PruneResult,
+    QueueItemDeliveryResult,
+    QueueItemOwner,
     RetryItemsResult,
     SessionQueueCountsByDestination,
     SessionQueueItem,
@@ -292,6 +294,27 @@ def get_queue_item_for_api(self, item_id: int) -> SessionQueueItem:
         """Gets a queue item in the response projection used by API reads."""
         return self.get_queue_item(item_id=item_id)
 
+    @abstractmethod
+    def get_queue_item_owner(self, item_id: int) -> QueueItemOwner:
+        """Gets a queue item's queue and owner without reading its session, for authorizing frequent requests."""
+        pass
+
+    @abstractmethod
+    def claim_queue_item_delivery(self, item_id: int, claimant: str) -> QueueItemDeliveryResult:
+        """Claims delivering a queue item's results for a claimant: `claimed` unless they were delivered or another
+        claimant's unexpired claim holds them."""
+        pass
+
+    @abstractmethod
+    def confirm_queue_item_delivery(self, item_id: int) -> QueueItemDeliveryResult:
+        """Marks a queue item's results delivered, whoever claimed them."""
+        pass
+
+    @abstractmethod
+    def release_queue_item_delivery(self, item_id: int, claimant: str) -> QueueItemDeliveryResult:
+        """Drops a claimant's claim of a queue item's undelivered results, if it holds one."""
+        pass
+
     @abstractmethod
     def get_queue_item_workflow_json(self, item_id: int) -> str | None:
         """Gets only the workflow metadata JSON for a queue item, if present."""

invokeai/app/services/session_queue/session_queue_common.py

@@ -2,7 +2,7 @@
 import json
 from collections.abc import Mapping
 from itertools import chain, product
-from typing import Any, Generator, Literal, Optional, TypeAlias, Union
+from typing import Any, Generator, Literal, NamedTuple, Optional, TypeAlias, Union
 
 from pydantic import (
     AliasChoices,
@@ -536,9 +536,51 @@ def validate_result(self) -> "EnqueueBatchReceipt":
         return self
 
 
+class RetriedQueueItem(BaseModel):
+    source_item_id: int = Field(description="The ID of the retried root queue item")
+    item_id: int = Field(description="The ID of the queue item that re-runs it")
+
+
 class RetryItemsResult(BaseModel):
     queue_id: str = Field(description="The ID of the queue")
     retried_item_ids: list[int] = Field(description="The IDs of the queue items that were retried")
+    retried_items: list[RetriedQueueItem] = Field(
+        default_factory=list, description="Each retried root queue item with the queue item that re-runs it"
+    )
+
+
+QueueItemDeliveryAction: TypeAlias = Literal["claim", "confirm", "release"]
+QueueItemDeliveryStatus: TypeAlias = Literal["claimed", "delivered", "held", "released"]
+
+
+class QueueItemDeliveryRequest(BaseModel):
+    """A change to who stages a queue item's results, so that only one browser does."""
+
+    action: QueueItemDeliveryAction = Field(
+        description="claim: take delivering the results; confirm: mark them delivered; release: give up a claim"
+    )
+    claimant: str = Field(
+        min_length=1, max_length=128, description="An id the client chooses for the browser tab or run claiming"
+    )
+
+
+class QueueItemOwner(NamedTuple):
+    """The queue a queue item is in and the account that owns it."""
+
+    queue_id: str
+    user_id: str
+
+
+class QueueItemDeliveryResult(BaseModel):
+    """Who stages a queue item's results after a delivery request."""
+
+    status: QueueItemDeliveryStatus = Field(
+        description="claimed: the claimant may deliver; delivered: the results were delivered; "
+        "held: another claimant is delivering them; released: the claimant holds no claim"
+    )
+    retry_after_ms: Optional[int] = Field(
+        default=None, description="When held, how long until the other claim lapses, in milliseconds"
+    )
 
 
 class ClearResult(BaseModel):