rw-lock test is not correct in ofstd/tests/tthread.cc, code use assumption that RWLockT1 (RWLockerT1) starts before RWLockT2 (RWLockerT2), this is not correct, pthread_create does not guarantee execution order and Mac OS X (El Capitan) time to time starts T2 before T1
easiest but not very correct fix is add OFStandard::milliSleep(wait_timeout) after line 302 and after line 424
ofthread test
Moderator: Moderator Team
-
- OFFIS DICOM Team
- Posts: 318
- Joined: Mon, 2014-03-03, 09:51
- Location: Oldenburg, Germany
Re: ofthread test
Thanks! We already noticed the problematic behavior of this unit test, but no one had the time to track it down so far. I have added your information to our bug tracker and assigned it to someone that already indicated his interest in looking at it. Let's see if this finally fixes those problems under OS X.
-
- OFFIS DICOM Team
- Posts: 318
- Joined: Mon, 2014-03-03, 09:51
- Location: Oldenburg, Germany
Re: ofthread test
It's fixed in this commit, thanks again for your input!
Re: ofthread test
np, thank you for the perfect framework,
btw if someone is interesting i (re-)implemented OFSemaphore on Mac OS El Capitan using c++11 mutex/condition_variable classes
not sure how to attach files to board now
btw if someone is interesting i (re-)implemented OFSemaphore on Mac OS El Capitan using c++11 mutex/condition_variable classes
not sure how to attach files to board now
-
- OFFIS DICOM Team
- Posts: 318
- Joined: Mon, 2014-03-03, 09:51
- Location: Oldenburg, Germany
Re: ofthread test
Sure, that should be helpful in general since the C++11 primitives are (finally!) platform independent. We'll need some preprocessor magic though, since DCMTK still needs to be compilable using pre C++11 compilers (yes, this hurts badly if you are used to C++11 features). Just send the patch to and we'll see if it can be integrated without problems.
Who is online
Users browsing this forum: No registered users and 1 guest