* fix: add proper sdk name to logs

* update CHANGELOG.md

* don't use deprecated `sentry_sdk_name()`

* update CHANGELOG.md

* add SENTRY_SDK_NAME fallback (fallforward?)

* fix typo

* initial traceparent support

* CHANGELOG.md update

* consolidate tests

* remove redundant value_len check

* check for exact length

* reuse header key comparison + extract parsing per header type

* static internal helper

* add size bounds + checks for headers

* only accept 00 or 01 sampling flag

* Apply suggestion from @supervacuus

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

* Apply suggestion from @supervacuus

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

* Apply suggestion from @supervacuus

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

* Apply suggestion from @supervacuus

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

* fix(traceparent): only support outgoing for now (#1406)

* cleanup

* remove feature of parsing incoming traceparent

* re-add len macros

* add tests to verify consistency of trace and span IDs across headers for transactions and spans

* use trace header sizes in header propagation

* update CHANGELOG.md

---------

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

* serialize uint64 values as numerical instead of string

* CHANGELOG.md

* construct unsigned int log params as sentry_value_t string type

* change format to expected unsigned long long int

* pass through the conversion character

* add tests for logs parameter conversion types

* less strict pointer argument check

…cov (#1419)

* Initial plan

* Fix crash test assertions to handle kcov's exit code behavior

Co-authored-by: vaind <6349682+vaind@users.noreply.github.com>

* Skip entire crash tests when running with kcov instead of conditional assertions

Co-authored-by: vaind <6349682+vaind@users.noreply.github.com>

* Only skip test_external_crash_reporter_http when running with kcov

Co-authored-by: vaind <6349682+vaind@users.noreply.github.com>

* Update tests/test_integration_http.py

* Update tests/test_integration_http.py

* Update tests/test_integration_http.py

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: vaind <6349682+vaind@users.noreply.github.com>

* Add logs-on-crash

* skip crashpad test on macos

* add straight-to-disk writing of log envelopes if crashing

* CHANGELOG.md

* restore thread join for normal shutdown

* directly take log item from httpserver log for crashpad test

* update CHANGELOG.md

* we know which envelope has the log

* we **don't** know which envelope has the log

* add more general request fetch from httpserver function

* add logs return value checks

* fix tsan global data race on logs validation

* move database-path cleanup into the cmake build cache fixture

this eliminates boilerplate noise in the tests but also reduces the chance of forgetting it in a test.

* in the crash-safe logs flush detach the batcher thread and don't wake it since we flush anyway.

* don't clean for unit-test crash_marker

* don't detach in crash-safe but spin-lock in its flusher to acquire flushing.

* add clean parameter to cmake fixture

* ensure logs flushing in crash handlers happens within the handler sync bounds.

* bound the spin lock in the logs flusher during crash-safe mode

* exclude crashed_last_run unit-test from cleaning database paths

* use an unsigned integer for the sleep_time of the crash-safe logs flusher spinlock

* provide cross-platform cpu_relax and replace syscall sleep with spinners.

* isolate `crash_marker` unit test and eliminate exceptions to database cleaning in the pytest configuration

* platform path special case in crash_marker unit test

* crashed_last_run can no longer assume that a database path exists

this means when it does its clearing a TEST_CHECK cannot assume that a directory even exists (in which case clear_crash_marker would actually fail versus the case were only the marker file didn't exist).

* clean up log statements

* bump CI python to 3.12

* fix CHANGELOG.md

---------

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

* feat: Auto-split build ID from version in sentry_library_info

When downstream SDKs modify sentry.h to include build metadata in the
version string (e.g., 0.11.3+20251016-9e31c9f-dirty), the embedded
library info now automatically extracts the build ID from the version.

Changes:
- Parse build metadata from SENTRY_VERSION_FULL if present
- Use base version (major.minor.patch) for SENTRY_VERSION field
- Use extracted build ID for BUILD field
- SENTRY_BUILD_ID cache variable still takes precedence
- Update template to use new SENTRY_EMBEDDED_VERSION and SENTRY_EMBEDDED_BUILD_ID
- Update tests to validate base version format
- Add test for build ID field

This ensures the embedded info format is:
SENTRY_VERSION:0.11.3;BUILD:20251016-9e31c9f-dirty
instead of:
SENTRY_VERSION:0.11.3+20251016-9e31c9f-dirty;BUILD:0.11.3+20251016-9e31c9f-dirty

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* refactor: Use SENTRY_VERSION_BASE directly instead of SENTRY_EMBEDDED_VERSION

Simplifies the code by reusing the existing SENTRY_VERSION_BASE variable
which already contains the major.minor.patch format without build metadata.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* refactor: Skip parsing if SENTRY_BUILD_ID is already set

Optimizes the logic to check SENTRY_BUILD_ID cache variable first.
Only attempts to extract build ID from version string if not explicitly provided.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* refactor: Simplify build ID logic with if-elseif-else chain

Cleaner and more readable conditional structure.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Use semver-compliant format for default build ID timestamp

According to semver.org spec, build metadata must contain only ASCII
alphanumerics and hyphens [0-9A-Za-z-]. Spaces are not allowed.

Changed from: "2025-10-17 10:59:00 UTC"
Changed to: "20251017-105900"

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* docs: Add changelog entry for automatic build ID extraction

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* feat: Add SENTRY_SDK_VERSION cache variable for downstream SDK override

Downstream SDKs can now override the SDK version at CMake configuration time
using -DSENTRY_SDK_VERSION="version+build-id". This allows setting the version
without modifying sentry.h.

When SENTRY_SDK_VERSION is set:
- The full version (with build metadata) is used for embedded library info
- Build ID is automatically extracted from the version string
- sentry.h uses the overridden version via compile definition
- Version parsing happens at CMake configuration time

Example usage:
cmake -DSENTRY_SDK_VERSION="0.11.3+20251016-9e31c9f-dirty" ...

Results in embedded info:
SENTRY_VERSION:0.11.3;BUILD:20251016-9e31c9f-dirty;...

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Check SENTRY_SDK_VERSION for empty string instead of truthiness

Aligns with the pattern used for SENTRY_SDK_NAME check.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Use SENTRY_SDK_VERSION directly for compile definitions instead of SENTRY_VERSION_FULL

* test: Add Python tests for SENTRY_SDK_VERSION override

Added two new integration tests:

1. test_sdk_version_override: Verifies that setting SENTRY_SDK_VERSION
   correctly separates version and build ID in the embedded library info.

2. test_sdk_version_override_with_explicit_build_id: Verifies that
   explicit SENTRY_BUILD_ID takes precedence over extracted build ID
   from the version string.

Both tests inspect the actual binary using the strings command to
validate the embedded information.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update regex to handle whitespace before SENTRY_SDK_VERSION define

The #ifndef guard adds leading whitespace to the #define line, so the
regex needs to handle optional leading whitespace and whitespace between
tokens.

This fixes the issue where CMake couldn't parse the version from sentry.h
when SENTRY_SDK_VERSION was not overridden.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update changelog entry for SENTRY_SDK_VERSION CMake cache variable to clarify build ID extraction

---------

Co-authored-by: Claude <noreply@anthropic.com>

Converts danger workflow from reusable workflow (v2) to composite
action (v3). This is a breaking change that requires restructuring
the job definition.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude <noreply@anthropic.com>

* add unit test

* only run on test_unit_transport

* add `sentry__bgworker_setname` again

* Update tests/unit/test_basic.c

* update CHANGELOG.md

* update crashpad

* update crashpad

* point crashpad submodule to getsentry branch

---------

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

* fix: make narrow utf-8 canonical across platforms

* fix unix path effects

* sentry__filewriter_byte_count now takes a const filewriter
* sentry__path_filename always returns a char.

* provide a platform-dependent accessor for the crashpad backend

* eliminate double-free sentry__path_from_str_owned.

* use correct format specifier for cli in debug logger for spawn

* Apply suggestion from @JoshuaMoelans

Co-authored-by: JoshuaMoelans <60878493+JoshuaMoelans@users.noreply.github.com>

* introduce `expect_failure` parameter to integration test run()

* reintroduce wide string comparisons in unit-tests

* update changelog

* clarify encoding in public header and clean up inline docs generally

* clarify encoding in the path header

* clarify remove behavior in the path header and fix windows implementation

* apply review feedback and various cleanups

* clean up

sentry__path_append_str()
sentry__path_absolute()
sentry__path_clone()

* update changelog

* format

* update breakpad

* fix "unknown command tag name"

* update breakpad

* update crashpad

* update changelog

* fix is_last_error_path_not_found() definition

* update breakpad to handler branch

* fixup changelog after master rebase

* add NULL check to sentry__string_clone_wstr

* clean up crash_marker path strin accessor

* add more `const` (#1422)

* Update include/sentry.h

Co-authored-by: JoshuaMoelans <60878493+JoshuaMoelans@users.noreply.github.com>

* Update include/sentry.h

Co-authored-by: JoshuaMoelans <60878493+JoshuaMoelans@users.noreply.github.com>

* limit system allocated `_wfullpath()` buffer to the `sentry__path_absolute()` scope

use `free()` internally in the function, and
clone the resulting wide-string at the end so we can be sure that the absolute path being passed around is not using system allocated buffer.

* document breaking change

* fix(logs): add `expect_failure` to logs_on_crash tests (#1421)

* add check for unexpected fail

* refactor logs tests to use expect_failure

* add missing `expect_failure` for logger tests

* fix copy mistake

* infer check `expect_failure` and remove from passed-in kwargs

* remove check from run_benchmark

* add check for `check` in kwargs

* Try to run the entire test-suite against a UTF-8 CWD

* realign with crashpad getsentry branch like master

* make the external crashreporter fixture wide-string aware on Windows.

* clean up external crashreporter fixture (fwprintf + SetConsoleOutputCP instead of explict wide to multibyte conversion)

* use wmain in the external crash reporter fixture only when building for _MSC_VER

* wait at the end of the WER integration tests that no WerFault.exe instances are still running

* clean up wide-to-multibyte string handling

---------

Co-authored-by: JoshuaMoelans <60878493+JoshuaMoelans@users.noreply.github.com>

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

…stack frame (#1427)

* Support optional usage of stack pointer for captured stack frame

---------

Co-authored-by: Mischan Toosarani-Hausberger <mischan@abovevacant.com>

* update bump-version.sh to handle sentry.h spaces

* uncomment

* regex independent of clang-format space amount

…on.sh (#1432)