Just re-raise the original exception, as-is

In Python `raise … from None` is an explicit request to raise the
exception but to suppress any exception chaining. (See:
https://docs.python.org/3/tutorial/errors.html#exception-chaining)

This means that if, when an ASGI exception crashes, an exception causes
another exception, that when the second exception (which has the first
chained to it) reaches this point in the stack, these lines will
re-raise it, but remove the chained exception. This can be confusing:
ASGI frameworks will normally log a stack trace of the crashing
application, but that stack trace will now be truncated, due to this
`from None` removing the chained exception.

The chained exception (which is the exception that kicked off the entire
stack unwind) is usually the more interesting one, when you're
debugging.

Neither of the two changed lines seem to have a reason for `… from
None`. Additionally, since they just re-raise the original exception,
simply "`raise`" suffices.

Fixes: #5624

Author: roy-work

sentry_sdk/integrations/asgi.py

@@ -183,7 +183,7 @@ async def _run_app(
 
             except Exception as exc:
                 self._capture_lifespan_exception(exc)
-                raise exc from None
+                raise
 
         _asgi_middleware_applied.set(True)
         try:
@@ -259,7 +259,7 @@ async def _sentry_wrapped_send(
                                 )
                         except Exception as exc:
                             self._capture_request_exception(exc)
-                            raise exc from None
+                            raise
         finally:
             _asgi_middleware_applied.set(False)