vSomeIP 3.5.0 #731
vSomeIP 3.5.0 #731
Conversation
fcmonteiro
commented
Jun 26, 2024
•
fcmonteiro
commented
- Load Policies Lazy Load
- Test - Processing SD messages with unknown type option
- ensure endpoints before deletion
- Improve "end of file" error handling
- Enable debouncing of events & selective events
- Revert "Test - Processing SD messages with unknown type"
- Logs added to points of failure on registration process
- One *.json to ignorem all
- Someip-tp remote address rework
- Fix crash in multicast_receive receive_cb
- Generate network_test configs directly to build
- Fix deadlock if binding of TCP client endpoint fails
- Added missing includes of iomanip to support compilation on Mint
- Cache not yet registered events
- Return true to make sure endpoints are deleted
- Byteorder implementation
- Reorder of prepare_stop method
- Allows applications in the same process using different security configurations
- Fix to not ignore stop offers when sd acceptance is not required
| @@ -109,6 +109,11 @@ if (ENABLE_SIGNAL_HANDLING) | |||
| set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -DVSOMEIP_ENABLE_SIGNAL_HANDLING") | |||
| endif () | |||
|
|
|||
| # Event caching | |||
| if (ENABLE_DEFAULT_EVENT_CACHING) | |||
| set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -DVSOMEIP_ENABLE_DEFAULT_EVENT_CACHING") | |||
There was a problem hiding this comment.
I'd recommend target_compile_definitions
| @@ -123,6 +115,7 @@ void tcp_client_endpoint_impl::restart(bool _force) { | |||
| << std::setw(4) << its_session << "]" | |||
| << " size: " << std::dec << q.first->size(); | |||
| } | |||
| self->sending_blocked_ = false; | |||
There was a problem hiding this comment.
I described in #668 that there's a defect where is_sending_ needs to get cleared. Clearing sending_blocked_ is not sufficient to fix that bug.
I suspect that both flags might need to get cleared during restart, but what is the rationale for setting sending_blocked_ false here? What was the bug?
There was a problem hiding this comment.
Hi @joeyoravec, as far as I know this was a "bug" where we were ignoring StopOffer messages (Fix to not ignore stop offers when sd acceptance is not required), however this pr was causing a lot of other problems and was reverted in 3.5.1. You can my check other pr in draft.
We will have to look at your issue #668.
