The struct + constexpr namespace approach caused the compiler to
materialize POLLING_UPDATE as a 4-byte symbol. An enum class with
uint32_t underlying type gives the same type safety but enum values
are true compile-time constants that never get materialized.
Move all set_timeout, cancel_timeout, set_interval, cancel_interval
overloads (const char*, uint32_t, std::string) from scheduler.cpp to
inline in scheduler.h. These are all trivial one-liner forwarding calls
to set_timer_common_ or cancel_item_ and benefit from inlining at
call sites rather than being separate symbols.
Move the 4 Scheduler InternalSchedulerID methods from .cpp to inline
in the header. They're trivial one-liners that extract .id and forward
to set_timer_common_/cancel_item_, so the compiler can fold them into
call sites instead of emitting separate function bodies.
Verifies that NUMERIC_ID_INTERNAL and NUMERIC_ID are completely
independent matching namespaces — same uint32_t value on the same
component does not collide. Tests that cancelling one type does not
affect the other, and that string names also don't cross-match.
Previously, PollingComponent used the string "update" and DelayAction
used "delay" as scheduler names. If a component subclassing
PollingComponent or using DelayAction also happened to use these same
strings for its own timers, the core timer would be silently cancelled
— a subtle bug that would be extremely hard to track down.
This introduces InternalSchedulerID, a type-safe wrapper that routes
through a new NUMERIC_ID_INTERNAL NameType. Since the scheduler
matches by (component, name_type, id, type), internal IDs can never
collide with component-level NUMERIC_ID or string-based names, even
if the underlying uint32_t values overlap.
Also migrates binary_sensor filters, MultiClickTrigger, and sensor
filters from string-based to uint32_t scheduler IDs. Each filter is
its own Component instance so IDs are scoped per-instance with no
collision risk.
- Update RECEIVE_CASES type hint from 2-tuple to 3-tuple to match
the actual stored shape (case, ifdef, message_name)
- Use consistent log format "%s: {}" for empty message overload
to match the "%s: %s" pattern of non-empty messages
Limit varint parsing to 10 bytes maximum (ceil(64/7)) to prevent
undefined behavior from shifting uint64_t by >= 64 bits when
processing malformed protobuf data with excessive continuation bytes.
On Windows, Path.resolve() treats '....' as parent traversal (403),
while on Unix it is a literal directory name that stays inside the
base directory (404).
hmac.compare_digest() on str inputs raises TypeError if either
contains non-ASCII characters. Encode both sides as UTF-8 bytes.
Add test with non-ASCII username to prevent regressions.
- Use auth.value().size() instead of strlen() to avoid rescanning
attacker-controlled header content.
- Iterate over digest_len (expected length) instead of max_len so a
long Authorization header cannot force extra work. The full-width
length XOR already rejects any length mismatch.
The test set mock_image.path to str, but FlashImage.path is a Path.
This masked a pre-existing bug where Path.endswith() doesn't exist.
Fix the mock to match the real type so as_posix() works correctly.
- Return 400 for empty or whitespace-only file_name to prevent the
idedata fallback from matching everything via empty-string suffix.
- Use image.path.as_posix().endswith() since FlashImage.path is a Path
object which does not have a string endswith method.
- Add parametrized test for empty/whitespace file name values.
- Change result accumulator from volatile uint8_t to volatile size_t
to prevent truncation bypass (e.g. digest_len + 256 XOR).
- Iterate over max(digest_len, provided_len) so trailing bytes in
either string are also compared.
- Refactor auth_settings fixture to extend dashboard_settings instead
of duplicating setup.
- Explicitly clear DISABLE_HA_AUTHENTICATION in HA add-on test to
prevent order-dependent flakiness.
J. Nick Koston