diff options
| author | Chris Cleeland <chris.cleeland@gmail.com> | 2009-10-29 21:34:59 +0000 |
|---|---|---|
| committer | Chris Cleeland <chris.cleeland@gmail.com> | 2009-10-29 21:34:59 +0000 |
| commit | 18374454ce4b490ec76c130d892af68e0237ee23 (patch) | |
| tree | ea43a3ebb987507e11ac54fc41cc0d1b999b3785 | |
| parent | 97312897257669cad7e52bdbcfef6c0b0adfe728 (diff) | |
| download | ATCD-18374454ce4b490ec76c130d892af68e0237ee23.tar.gz | |
Thu Oct 29 21:25:49 UTC 2009 Chris Cleeland <cleeland@ociweb.com>
* tests/Message_Queue_Test.cpp:
Reverted change from Tue Oct 27 08:12:25 UTC 2009
William R. Otte <wotte@dre.vanderbilt.edu>. The change
was causing core dumps on some platforms due to treating
the thread ID as a pointer and dereferencing that pointer.
Better to live with compiler warnings than crashes.
| -rw-r--r-- | ACE/ChangeLog | 10 | ||||
| -rw-r--r-- | ACE/tests/Message_Queue_Test.cpp | 16 |
2 files changed, 12 insertions, 14 deletions
diff --git a/ACE/ChangeLog b/ACE/ChangeLog index f3960d5c1ba..da9430ca971 100644 --- a/ACE/ChangeLog +++ b/ACE/ChangeLog @@ -1,3 +1,13 @@ +Thu Oct 29 21:25:49 UTC 2009 Chris Cleeland <cleeland@ociweb.com> + + * tests/Message_Queue_Test.cpp: + + Reverted change from Tue Oct 27 08:12:25 UTC 2009 + William R. Otte <wotte@dre.vanderbilt.edu>. The change + was causing core dumps on some platforms due to treating + the thread ID as a pointer and dereferencing that pointer. + Better to live with compiler warnings than crashes. + Thu Oct 29 21:04:27 UTC 2009 Olli Savia <ops@iki.fi> * ace/README: diff --git a/ACE/tests/Message_Queue_Test.cpp b/ACE/tests/Message_Queue_Test.cpp index 3c5c0c467be..a0192186797 100644 --- a/ACE/tests/Message_Queue_Test.cpp +++ b/ACE/tests/Message_Queue_Test.cpp @@ -123,13 +123,7 @@ Counting_Test_Producer::svc (void) // correct. // Also, to be sure there's not just 1 producer and 1 consumer pinging // back and forth, make the producers randomly delay between blocks. - - // Seeding the random number generator like this really isn't a good idea, - // due to the wild differences between platforms with respect to the definition - // of ACE_thread_t. These pointer gymnastics are intended to suppress warnings - // that we really don't care about. - void *thr_id = reinterpret_cast <void *> (ACE_Thread::self ()); - ACE_OS::srand (*reinterpret_cast <u_int *> (thr_id)); + ACE_OS::srand ((u_int)ACE_Thread::self ()); int multiple = ACE_OS::rand () % 10; int delay_ms = (ACE_OS::rand () % 10) / 2; // The delay usually causes the test to time out in the automated @@ -219,13 +213,7 @@ Counting_Test_Consumer::svc (void) // a calculated number of blocks then stop; the test checker will determine // if the number consumed plus the number remaining is correct for the // number produced. - - // Seeding the random number generator like this really isn't a good idea, - // due to the wild differences between platforms with respect to the definition - // of ACE_thread_t. These pointer gymnastics are intended to suppress warnings - // that we really don't care about. - void *thr_id = reinterpret_cast <void *> (ACE_Thread::self ()); - ACE_OS::srand (*reinterpret_cast <u_int *> (thr_id)); + ACE_OS::srand ((u_int)ACE_Thread::self ()); int multiple = ACE_OS::rand () % 10; int delay_ms = ACE_OS::rand () % 10; // The delay usually causes the test to time out in the automated |