samiandClaude Sonnet 5
5d81b4cdae
vdm/ids.hpp — TaskId, an opaque engine handle (DAEMON keeps the wire UUID <-> TaskId map; the engine never sees the UUID). segment/segmenter — per-download range management (docs/04 §3). Initial lazy split; assign_slot() splits the largest remaining range when the budget grants a slot; on_complete(may_steal) either *steals* the second half of the largest remaining range for the same worker (slot-neutral) or returns nullopt so the caller *yields* the slot (ADR 0011 A1); on_failed() returns requeue only on the 3rd consecutive connection error with a mirror present — the remaining range is orphaned and re-split. Non-resumable or unknown-size => exactly 1 segment; never split below min_segment_bytes (1 MiB). Resume ctor rebuilds from a persisted table (falls back to a fresh layout if it doesn't tile [0,total)). One mutex == "the task lock"; segment fields are std::atomic and the store is a std::deque so a steal's append never moves a worker's record. segment/budget — the global allocator (ADR 0011). Owns exactly one ceiling (maxActiveSegments) and min-1-before-seconds fairness: a two-pass allocation (guarantee pass gives every wanting task 1 slot in DAEMON's priority order, then a growth pass round-robins the rest up to each task's effective cap = min(per_task_cap, host cap, 1 if non-resumable)), recomputed from scratch on every edge so a live set_max_active_segments cut naturally yields the excess lowest-priority-first, never a mid-segment kill. DAEMON-facing surface exactly as promised in daemon/docs/core-requests-m1.md / ADR 0011: budget(), segments_active(), starved_tasks(), starved_since(), set_max_active_segments (drain), set_host_segment_cap, set_task_order, on_budget_changed (a jthread coalesces at <=4 Hz; the tasks_starved 0<->nonzero edge fires immediately). Callbacks are copied out and run after the lock is dropped. Tests: segmenter split/steal/requeue/resume math + a concurrent steal-and-advance run; budget min-1 under a tight budget, round-robin growth, host-cap and non-resumable clamps, live-lower shedding lowest-priority-first, starvation below the task count, starved-edge notification, and a concurrent set_want hammer. Green under ASan/UBSan; the steal path and the budget green under TSan. Co-Authored-By: Claude Sonnet 5 <[email protected]> Claude-Session: https://claude.ai/code/session_01HPPSGhiArbvQgwC2DNiURS
libveloxcore — public API
Status: M1 in progress. Only util/ is landed. The download-facing API
(DownloadSpec, DownloadTask, probe, typed callbacks) arrives with later stages and
is reviewed by DAEMON before M2 (AGENT-CORE DoD).
Layering (CLAUDE.md §3): this library knows nothing about JSON, SQL, Qt, or RPC. Input is
a spec value; output is bytes on disk plus typed callbacks. DAEMON projects engine state
onto the wire contract's TaskSummary / TaskDetail / events — see
core/docs/proto-requests-m1.md for the shapes that projection needs frozen.
Every header under core/include/vdm/ compiles standalone (-Wall -Wextra -Wpedantic -Werror, C++23). Clean under ASan/UBSan and TSan.
util/ — foundations
vdm/util/error.hpp
enum class Error — the engine-wide failure taxonomy (network / HTTP / content / local
I/O / metadata / probe / retry / internal). This is CORE's own vocabulary; it is not
a wire type. error_name(Error) gives a stable snake_case string; is_retryable(Error)
is the advisory retry hint the task policy consults.
struct ErrorInfo { Error code; std::string context; int http_status; bool retryable; Error cause; } — the payload carried by every failed Result. .to_string() renders
"<name>: <context> (HTTP <n>)".
vdm/util/result.hpp
Result<T> — return-based error channel, a thin wrapper over
std::expected<T, ErrorInfo>. Errors are returned, never thrown, on anything that
runs during a transfer.
Result<int> r = 42;/Result<int> r = Err{Error::timeout, "..."};/Result<T> r = Error::not_found;r.has_value(),explicit operator bool,r.value()/*r/r->,r.error(),r.code(),r.value_or(x)- monadic
and_then/transform/transform_error(forward tostd::expected) Result<void>specialization;vdm::ok()success sentinelVDM_TRY(expr)— return the error ifexprfailedVDM_TRY_ASSIGN(auto x, expr)— bind the value or return the error
vdm/util/bytes.hpp
Byte / ByteSpan / ConstByteSpan aliases; as_bytes(string_view) /
as_chars(span). Little-endian fixed-width codec load_le<T> / store_le<T> and a
bounds-checked sequential ByteReader (.u8/.u16/.u32/.u64, .raw(n), .lp_string(),
.overran()). Built for the .veloxpart.meta reader and the 4-byte NM framing; every
read is bounds-checked and latches on overrun (reader-first, fuzz-ready).
vdm/util/event_bus.hpp
EventBus — typed, thread-safe in-process pub/sub. subscribe<E>(fn) -> Token,
publish<E>(ev) (synchronous, calling thread, registration order), unsubscribe(Token),
and RAII subscribe_scoped<E> returning a Subscription. Handlers may (un)subscribe or
publish during dispatch. Handlers must not throw. Not a hot-path structure — progress is
coalesced to ≤4 Hz upstream.
vdm/util/thread_pool.hpp
ThreadPool — fixed-size std::jthread pool for bounded off-loop work (hashing, fsync
batches, DNS pre-resolve). submit(fn, args...) -> std::future<R>; propagates exceptions
through the future; drains already-queued tasks on destruction. Not the transfer
loop — net/ will own one curl_multi per dedicated worker.
vdm/util/log.hpp
Sink interface — core does no I/O itself. LogSink abstract base; DAEMON installs one
via set_log_sink(), default discards. CallbackSink adapter (with a min-level filter).
VDM_LOG_{TRACE,DEBUG,INFO,WARN,ERROR}(category, fmt, args...) — std::format syntax,
only formatted when a sink is installed and wants the level.