Daily edition

Sep 3, 2026

The edition exactly as it was published. Nothing here is re-selected on a later reading.

Sep 4, 20264 reads · 138 merged PRs screened

language design

rust-lang/rust · #162120

Files changed →View PR ↗

Introduce `PerOwnerLoweringState`Rust · 241 + / 250

Introduces 5 new declarations in compiler/rustc_ast_lowering/src/lib.rs.

Adds new language design rather than adjusting what was there. Read the linked PR for the surrounding test context.

compiler/rustc_ast_lowering/src/lib.rs · 10 files

@@ -131,36 +131,155 @@ pub(crate) mod re_lowering {             ctx: &mut LoweringContext<'a, 'hir>,             op: impl FnOnce(&mut LoweringContext<'a, 'hir>) -> TRes,         ) -> TRes {-            assert!(!ctx.relowering_checker.can_relower, "reentrant relowering is not supported");+            assert!(+                !ctx.curr_owner.relowering_checker.can_relower,+                "reentrant relowering is not supported"+            ); -            ctx.relowering_checker.can_relower = true;+            ctx.curr_owner.relowering_checker.can_relower = true;              let res = op(ctx); -            ctx.relowering_checker.can_relower = false;+            ctx.curr_owner.relowering_checker.can_relower = false;              res         }     } } -struct LoweringContext<'a, 'hir> {-    tcx: TyCtxt<'hir>,-    resolver: &'a ResolverAstLowering<'hir>,-    current_disambiguator: PerParentDisambiguatorState,+struct PerOwnerLoweringState<'a, 'hir> {+    // -- Identity --+    owner: &'a PerOwnerResolverData<'hir>,+    owner_id: hir::OwnerId,+    disambiguator: PerParentDisambiguatorState, -    /// Used to allocate HIR nodes.

core implementation

microsoft/vscode · #334378

Files changed →View PR ↗

agentHost: classify inline chat telemetryTypeScript · 48 + / 21

Introduces 2 new declarations in src/vs/platform/agentHost/node/agentHostTelemetryReporter.ts.

Adds new core implementation rather than adjusting what was there. Tests changed with it, with code of their own.

src/vs/platform/agentHost/node/agentHostTelemetryReporter.ts · 7 files

@@ -24,16 +24,18 @@ import { AgentHostClientConnectionKind, AgentHostLaunchKind, AgentHostTransportK export type AgentHostUserMessageSentSource = 'direct' | 'queued';  /**- * Who produced the message that started a turn. Extends the protocol's- * {@link MessageKind} with `agentMerge`: Agent Merge drives its repair turns- * with a host-generated message that carries the `systemNotification` origin,- * and reporting those under their own value keeps automated merge work- * separable from turns a person or an agent asked for.+ * The message origin used for telemetry. Extends the protocol's+ * {@link MessageKind} with host-owned classifications for Agent Merge repair+ * turns and VS Code sessions marked ephemeral, which telemetry reports as+ * `inline`.  */-export type AgentHostMessageOriginTelemetryKind = MessageKind | 'agentMerge';+export type AgentHostMessageOriginTelemetryKind = MessageKind | 'agentMerge' | 'inline'; -/** Classifies the actor that produced a turn's message for telemetry. */-export function getMessageOriginTelemetryKind(message: Message): AgentHostMessageOriginTelemetryKind {+/** Classifies a turn's message origin, including host-owned session classifications. */+export function getMessageOriginTelemetryKind(message: Message, isEphemeralSession: boolean): AgentHostMessageOriginTelemetryKind {+	if (isEphemeralSession) {+		return 'inline';+	} 	// The marker only counts on the origin the host stamps it with, so a client 	// cannot dress a user message up as automated merge work. 	if (message.origin.kind === MessageKind.SystemNotification && isAgentMergeMessage(message)) {

systems design

kubernetes/kubernetes · #141533

Files changed →View PR ↗

discovery: graduate declarative validation to stableGo · 25 + / 52

Reworks 25 lines of existing logic in pkg/apis/discovery/validation/validation.go.

Changes how existing systems design behaves. Tests changed with it, with code of their own.

pkg/apis/discovery/validation/validation.go · 9 files

@@ -202,18 +187,6 @@ func validatePorts(endpointPorts []discovery.EndpointPort, fldPath *field.Path) 	return allErrs } -func validateAddressType(addressType discovery.AddressType) field.ErrorList {-	allErrs := field.ErrorList{}--	if addressType == "" {-		allErrs = append(allErrs, field.Required(field.NewPath("addressType"), "").MarkCoveredByDeclarative())-	} else if !supportedAddressTypes.Has(addressType) {-		allErrs = append(allErrs, field.NotSupported(field.NewPath("addressType"), addressType, sets.List(supportedAddressTypes)).MarkCoveredByDeclarative())-	}--	return allErrs-}- func validateHints(endpointHints *discovery.EndpointHints, fldPath *field.Path) field.ErrorList { 	allErrs := field.ErrorList{} 

core implementation

huggingface/transformers · #48436

Files changed →View PR ↗

[`Qwen 3.5 Moe`] Fix decoratorsPython · 12 + / 2

Reworks 9 lines of existing logic in src/transformers/models/qwen3_5_moe/modular_qwen3_5_moe.py.

Changes how existing core implementation behaves. Read the linked PR for the surrounding test context.

src/transformers/models/qwen3_5_moe/modular_qwen3_5_moe.py · 2 files

@@ -31,6 +31,10 @@     Qwen3_5TextRotaryEmbedding,     Qwen3_5VisionModel,     Qwen3_5VisionRotaryEmbedding,+    causal_conv1d_fn,+    causal_conv1d_update,+    torch_chunk_gated_delta_rule,+    torch_recurrent_gated_delta_rule, ) from ..qwen3_next.configuration_qwen3_next import Qwen3NextConfig from ..qwen3_next.modeling_qwen3_next import (

Read today’s edition →