Hallo, if nobody objects I plan to commit the attached patch in a couple of days. It just extends the existing workaround to the latest compiler release. Cheers, Nicola Musatti ******Index: read_write_mutex.hpp =================================================================== RCS file: /cvsroot/boost/boost/boost/thread/read_write_mutex.hpp,v retrieving revision 1.9 diff -d -u -r1.9 read_write_mutex.hpp --- read_write_mutex.hpp 11 Jul 2005 15:06:07 -0000 1.9 +++ read_write_mutex.hpp 2 May 2006 20:59:52 -0000 @@ -53,7 +53,7 @@ typedef detail::thread::scoped_timed_lock<Mutex> scoped_timed_lock; read_write_mutex_impl(read_write_scheduling_policy::read_write_scheduling_policy_enum sp); -#if !BOOST_WORKAROUND(__BORLANDC__,<= 0x564) +#if !BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(582)) ~read_write_mutex_impl(); #endif ******Index: read_write_mutex.cpp =================================================================== RCS file: /cvsroot/boost/boost/libs/thread/src/read_write_mutex.cpp,v retrieving revision 1.23 diff -d -u -r1.23 read_write_mutex.cpp --- read_write_mutex.cpp 27 Aug 2005 10:25:15 -0000 1.23 +++ read_write_mutex.cpp 2 May 2006 20:59:40 -0000 @@ -379,7 +379,7 @@ , m_readers_next(true) {} -#if !BOOST_WORKAROUND(__BORLANDC__, <= 0x564) +#if !BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(582)) template<typename Mutex> read_write_mutex_impl<Mutex>::~read_write_mutex_impl() {
Nicola Musatti wrote:
if nobody objects I plan to commit the attached patch in a couple of days. It just extends the existing workaround to the latest compiler release.
I cannot see a reason why not to do so. May I assume you have tested, that the workaround still is needed? I have no current borland compiler around to verify this. I would like to ask you also to put a short comment into the source as why the work around is needed at all. Roland
Roland Schwarz wrote:
Nicola Musatti wrote:
if nobody objects I plan to commit the attached patch in a couple of days. It just extends the existing workaround to the latest compiler release.
I cannot see a reason why not to do so. May I assume you have tested, that the workaround still is needed? I have no current borland compiler around to verify this.
Thank you. I did test it. Actually I decided to reenable this workaround due to the test_read-write_mutex test failing with the latest Borland compiler which is passing with bcc 5.6.4.
I would like to ask you also to put a short comment into the source as why the work around is needed at all.
I will. Cheers, Nicola Musatti
participants (2)
-
Nicola Musatti -
Roland Schwarz