* split up test_integration_http.py

* cleanup unnecessary `auth_header`
- already tested in `test_integration_http`, so no need to also have it for each of the other integration tests.

Co-Authored-By: Claude <[email protected]>

---------

Co-authored-by: Claude <[email protected]>

MiniDumpWriteDump can deadlock on Windows due to loader lock
contention (see #1501). Add a 5-minute timeout and retry up to 3
times to prevent the test from hanging indefinitely in CI.

Co-authored-by: Claude Opus 4.5 <[email protected]>

* feat: add sentry__path_rename()

Partial cherry pick from commit 2135f0f

* test: add unit tests for sentry__path_rename()

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix: add null guards before memcmp in path_rename test

Co-Authored-By: Claude Opus 4.6 <[email protected]>

---------

Co-authored-by: JoshuaMoelans <[email protected]>
Co-authored-by: Claude Opus 4.6 <[email protected]>

Co-authored-by: Claude Opus 4.6 <[email protected]>

Co-authored-by: Claude Opus 4.6 <[email protected]>

Co-authored-by: Claude Opus 4.5 <[email protected]>
Co-authored-by: Claude Opus 4.6 <[email protected]>
Co-authored-by: JoshuaMoelans <[email protected]>

Co-authored-by: Claude Opus 4.6 <[email protected]>

Co-authored-by: Claude Opus 4.5 <[email protected]>
Co-authored-by: Claude Opus 4.6 <[email protected]>

…tforms (#1517)

* fix(tests): skip cache tests on NX and PS platforms

* Skip path_rename test for NX

* test(logs): add reinit deadlock test

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* test(metrics): add reinit deadlock test

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix(batcher): eliminate g_options_lock from flush path to prevent deadlock

When sentry_init() is called while a batcher thread is mid-flush,
a deadlock occurs: the main thread holds g_options_lock and waits
for the batcher thread to join, while the batcher thread tries to
acquire g_options_lock via SENTRY_WITH_OPTIONS during flush.

Store the options pointer in the batcher at startup and use it
directly during flush instead of going through SENTRY_WITH_OPTIONS.
Add sentry__envelope_new_with_dsn() to create envelopes without
locking, and sentry__options_get_user_consent() for lock-free
consent checks.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* ref(batcher): store individual fields instead of options pointer

Replace the raw options pointer with individual fields (dsn,
transport, run, user_consent) to avoid unsynchronized access to
options members from the batcher thread. The dsn is incref'd,
and user_consent is a pointer to the atomic field (NULL when
consent is not required).

Revert sentry__options_get_user_consent since it is no longer
needed.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* docs: add changelog entry for batcher deadlock fix

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix(envelope): hold DSN ref across sentry__envelope_new_with_dsn call

The DSN pointer was extracted under g_options_lock but used after
the lock was released, racing with sentry_close freeing it.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix(test): enable metrics instead of logs in metrics_reinit test

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix(batcher): release DSN ref when thread spawn fails

Co-Authored-By: Claude Opus 4.6 <[email protected]>

---------

Co-authored-by: Claude Opus 4.6 <[email protected]>