Joel Sherrill commented: https://gitlab.rtems.org/rtems/rtos/rtems/-/issues/3791#note_143505 I am not able to review the MR tonight but here is some feedback: 1. I think this approach gives us a path to have the blocking discipline selected on a per instance basis which is a good enough starting point. We can have it default to FIFO and file an issue for 8.1 to change the default from FIFO to priority. This will require adding something to the release notes but gives us a path to change the behavior with plenty of notice. 2. Everything is compiled from source with RTEMS so we side-step those issues. Note _mq_ispriority_ should be _mq_ispriority_np_. The "_np" is non-portable and indicates it is an extension. That's a POSIX convention -- not RTEMS. 3. I don't think we need a _confdefs.h_ mechanism yet. If @gedare is OK with it, I am thinking that making the current default attributes have a weak name of something like _rtems_posix_message_queue_default_attributes_ and documenting that the user can provide their own override version of the structure is a good solution. This pattern could be applied to all POSIX objects with attribute structures. @gedare How does all this sound to you? -- View it on GitLab: https://gitlab.rtems.org/rtems/rtos/rtems/-/issues/3791#note_143505 You're receiving this email because of your account on gitlab.rtems.org.
_______________________________________________ bugs mailing list [email protected] http://lists.rtems.org/mailman/listinfo/bugs
