Hi, I've noticed that my Boost.Log tests on MSVC are failing massively. The build log hints that Boost.Thread might be the culprit, and its tests are also failing on MSVC with the following error: ..\boost/thread/future.hpp(1127) : error C2248: 'boost::unique_lock<Mutex>::operator =' : cannot access private member declared in class 'boost::unique_lock<Mutex>' with [ Mutex=boost::mutex ] ..\boost/thread/lock_types.hpp(110) : see declaration of 'boost::unique_lock<Mutex>::operator =' with [ Mutex=boost::mutex ] This is happening on release branch. Could this be fixed?
Le 15/06/13 12:23, Andrey Semashev a écrit :
Hi,
I've noticed that my Boost.Log tests on MSVC are failing massively. The build log hints that Boost.Thread might be the culprit, and its tests are also failing on MSVC with the following error:
..\boost/thread/future.hpp(1127) : error C2248: 'boost::unique_lock<Mutex>::operator =' : cannot access private member declared in class 'boost::unique_lock<Mutex>' with [ Mutex=boost::mutex ] ..\boost/thread/lock_types.hpp(110) : see declaration of 'boost::unique_lock<Mutex>::operator =' with [ Mutex=boost::mutex ]
This is happening on release branch. Could this be fixed?
Hi, I have no access to a windows machine now. There were this change set https://svn.boost.org/trac/boost/changeset/84750 that could be related. Please could you try to revert this and let me know if this is the source of the problems? Index: branches/release/boost/thread/detail/delete.hpp =================================================================== --- a/branches/release/boost/thread/detail/delete.hpp +++ b/branches/release/boost/thread/detail/delete.hpp @@ -24,4 +24,15 @@ #else // BOOST_NO_CXX11_DELETED_FUNCTIONS +#if defined(BOOST_MSVC) +#define BOOST_THREAD_DELETE_COPY_CTOR(CLASS) \ + private: \ + CLASS(CLASS const&); \ + public: + +#define BOOST_THREAD_DELETE_COPY_ASSIGN(CLASS) \ + private: \ + CLASS& operator=(CLASS const&); \ + public: +#else #define BOOST_THREAD_DELETE_COPY_CTOR(CLASS) \ private: \ @@ -33,4 +44,5 @@ CLASS& operator=(CLASS&); \ public: +#endif #endif // BOOST_NO_CXX11_DELETED_FUNCTIONS Best, Vicente
Le 15/06/13 13:02, Vicente J. Botet Escriba a écrit :
Le 15/06/13 12:23, Andrey Semashev a écrit :
Hi,
I've noticed that my Boost.Log tests on MSVC are failing massively. The build log hints that Boost.Thread might be the culprit, and its tests are also failing on MSVC with the following error:
..\boost/thread/future.hpp(1127) : error C2248: 'boost::unique_lock<Mutex>::operator =' : cannot access private member declared in class 'boost::unique_lock<Mutex>' with [ Mutex=boost::mutex ] ..\boost/thread/lock_types.hpp(110) : see declaration of 'boost::unique_lock<Mutex>::operator =' with [ Mutex=boost::mutex ]
This is happening on release branch. Could this be fixed?
Hi,
I have no access to a windows machine now.
There were this change set https://svn.boost.org/trac/boost/changeset/84750 that could be related.
Please could you try to revert this and let me know if this is the source of the problems?
BTW, I see that there is an error for versions 8 and 9 but no error for version 10 and 11. With which versions are you finding the error? Best, Vicente
On Saturday 15 June 2013 13:08:04 Vicente J. Botet Escriba wrote:
Le 15/06/13 13:02, Vicente J. Botet Escriba a écrit :
Le 15/06/13 12:23, Andrey Semashev a écrit :
Hi,
I've noticed that my Boost.Log tests on MSVC are failing massively. The build log hints that Boost.Thread might be the culprit, and its tests are also failing on MSVC with the following error:
I have no access to a windows machine now.
There were this change set https://svn.boost.org/trac/boost/changeset/84750 that could be related.
Please could you try to revert this and let me know if this is the source of the problems?
BTW, I see that there is an error for versions 8 and 9 but no error for version 10 and 11. With which versions are you finding the error?
I didn't try to reproduce the error locally yet, I'm just looking at the test results table. I will be able to try it myself a bit later (this evening, probably).
Le 15/06/13 13:08, Vicente J. Botet Escriba a écrit :
Le 15/06/13 13:02, Vicente J. Botet Escriba a écrit :
Le 15/06/13 12:23, Andrey Semashev a écrit :
Hi,
I've noticed that my Boost.Log tests on MSVC are failing massively. The build log hints that Boost.Thread might be the culprit, and its tests are also failing on MSVC with the following error:
..\boost/thread/future.hpp(1127) : error C2248: 'boost::unique_lock<Mutex>::operator =' : cannot access private member declared in class 'boost::unique_lock<Mutex>' with [ Mutex=boost::mutex ] ..\boost/thread/lock_types.hpp(110) : see declaration of 'boost::unique_lock<Mutex>::operator =' with [ Mutex=boost::mutex ]
This is happening on release branch. Could this be fixed?
Hi,
I have no access to a windows machine now.
There were this change set https://svn.boost.org/trac/boost/changeset/84750 that could be related.
Please could you try to revert this and let me know if this is the source of the problems?
BTW, I see that there is an error for versions 8 and 9 but no error for version 10 and 11. With which versions are you finding the error?
Please, could you try by changing #if defined(BOOST_MSVC) by #if defined(BOOST_MSVC) && _MSC_VER >= 1600 Vicente
On Sat, Jun 15, 2013 at 3:17 PM, Vicente J. Botet Escriba < vicente.botet@wanadoo.fr> wrote:
Please, could you try by changing
#if defined(BOOST_MSVC)
by
#if defined(BOOST_MSVC) && _MSC_VER >= 1600
Yes, that seem to fix the problem in my local runs. Will you commit it?
On Sun, Jun 16, 2013 at 9:57 PM, Andrey Semashev
On Sat, Jun 15, 2013 at 3:17 PM, Vicente J. Botet Escriba < vicente.botet@wanadoo.fr> wrote:
Please, could you try by changing
#if defined(BOOST_MSVC)
by
#if defined(BOOST_MSVC) && _MSC_VER >= 1600
Yes, that seem to fix the problem in my local runs. Will you commit it?
BTW, you may be interested in the new BOOST_DELETED_FUNCTION macro I added to Boost.Config recently. Though I didn't inspect Boost.Thread code closely to see if it's suitable.
Le 16/06/13 19:57, Andrey Semashev a écrit :
On Sat, Jun 15, 2013 at 3:17 PM, Vicente J. Botet Escriba < vicente.botet@wanadoo.fr> wrote:
Please, could you try by changing
#if defined(BOOST_MSVC)
by
#if defined(BOOST_MSVC) && _MSC_VER >= 1600
Yes, that seem to fix the problem in my local runs. Will you commit it?
Committed revision 84807. I'll merge it once I see the errors disappear on trunk. Vicente
I see that there is an error for versions 8 and 9 but no error for version 10 and 11.
I tried running the log tests locally with VC9 and they all pass for me here, I note that the test runner is using an unpatched (no service pack maybe?) version of VC9 - his is version 150021022, mine is 150030729. Perhaps we should just encourage him to upgrade? John.
On 6/15/2013 11:53 AM, John Maddock wrote:
I see that there is an error for versions 8 and 9 but no error for version 10 and 11.
I tried running the log tests locally with VC9 and they all pass for me here, I note that the test runner is using an unpatched (no service pack maybe?) version of VC9 - his is version 150021022, mine is 150030729. Perhaps we should just encourage him to upgrade?
I already mentioned that fact to him, in relation to some TTI failures which pass in my own tests. He acknowledged it but did not say whether he would upgrade to SP1 or not.
On Sun, Jun 16, 2013 at 9:29 PM, Edward Diener
On 6/15/2013 11:53 AM, John Maddock wrote:
I see that there is an error for versions 8 and 9 but no error for
version 10 and 11.
I tried running the log tests locally with VC9 and they all pass for me here, I note that the test runner is using an unpatched (no service pack maybe?) version of VC9 - his is version 150021022, mine is 150030729. Perhaps we should just encourage him to upgrade?
I already mentioned that fact to him, in relation to some TTI failures which pass in my own tests. He acknowledged it but did not say whether he would upgrade to SP1 or not.
I reproduced the problem locally with VC8 (14.00.50727.762 for 80x86) and VC9 (15.00.30729.01 for 80x86) - all 32-bit. Also, at least some time ago the tests were passing for this tester.
participants (4)
-
Andrey Semashev -
Edward Diener -
John Maddock -
Vicente J. Botet Escriba