From: Konstantin L. <Kon...@ma...> - 2007-07-10 07:35:17
|
Hello All! I have finished build for msvc 8.0, MinGW32 (gcc 3.4.2), borland 5.6.4 and gcc 4.1.1 :) Everything seams to work. This is not a simple bugfix release because we added factories into codebase. So may be we should increase minor version? Post here because have no response from waffel. With best regards, Konstantin Litvinenko. |
From: Fathi B. <fb...@fr...> - 2007-07-26 20:03:18
|
hi all, nice to hear that there's an upcoming log4cpp new release. i received a bug report on the debian about a threading issue: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=432729 could someone take a look (waffel ? other?) at the submitted patch and commit it if it is ok before the official release. --- ./log4cpp-0.3.5-rc3/src/Appender.cpp 2002-10-27 02:48:52.000000000 +0100 +++ ./log4cpp-0.3.5rc1/src/Appender.cpp 2004-10-26 22:08:28.000000000 +0200 @@ -11,14 +11,11 @@ #include <log4cpp/Appender.hh> namespace log4cpp { - Appender::AppenderMap* Appender::_allAppenders; threading::Mutex Appender::_appenderMapMutex; /* assume _appenderMapMutex locked */ Appender::AppenderMap& Appender::_getAllAppenders() { - if (!_allAppenders) - _allAppenders = new Appender::AppenderMap(); - + static Appender::AppenderMap* _allAppenders = new Appender::AppenderMap(); return *_allAppenders; } cheers, Fathi |
From: Konstantin L. <Kon...@ma...> - 2007-07-27 07:40:46
|
Hello, Fathi! You wrote to <log...@li...> on Thu, 26 Jul 2007 22:03:15 +0200: FB> nice to hear that there's an upcoming log4cpp new release. FB> i received a bug report on the debian about a threading issue: FB> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=432729 FB> could someone take a look (waffel ? other?) at the submitted patch and FB> commit it if it is ok before the official release. FB> namespace log4cpp { FB> - Appender::AppenderMap* Appender::_allAppenders; FB> threading::Mutex Appender::_appenderMapMutex; FB> /* assume _appenderMapMutex locked */ FB> Appender::AppenderMap& Appender::_getAllAppenders() { FB> - if (!_allAppenders) FB> - _allAppenders = new Appender::AppenderMap(); FB> - FB> + static Appender::AppenderMap* _allAppenders = new FB> Appender::AppenderMap(); FB> return *_allAppenders; FB> } I don't see why folks thinks that it is threading issue, but I see very, very strange bug with line Appender::AppenderMap* Appender::_allAppenders; How it's works in current release I can't understand. This line should be Appender::AppenderMap* Appender::_allAppenders = 0; I have looked the bug report and don't think it is threading issue too. I think we have some problems with NDC. I will look on it closely on weekend. With best regards, Konstantin Litvinenko. |
From: <fb...@fr...> - 2007-07-27 09:56:53
|
Hi konstantin, all, > I have looked the bug report and don't think it is threading issue too. I > think we have some problems with NDC. I will look on it closely on weekend. thanks ! could you take a look to: https://spaces.internet2.edu/display/SHIB/Log4Cpp > During the course of Shibboleth development, the original author orphaned > the library and a set of patches was locally applied to produce a custom > package to correct some thread safety and shutdown issues in the code. The > project has since revived but a sync-up has not yet occurred. You MUST use > the Internet2-supplied version or you will not have a reliable package. > The locally produced source package is available at > http://shibboleth.internet2.edu/downloads/log4cpp-0.3.5rc1.tar.gz i will ping shibboleth guys, it could be nice to merge the changes and do the sync. cheers, Fathi |
From: Konstantin L. <Kon...@ma...> - 2007-07-27 07:40:48
|
Hello, Fathi! You wrote to <log...@li...> on Thu, 26 Jul 2007 22:03:15 +0200: FB> nice to hear that there's an upcoming log4cpp new release. FB> i received a bug report on the debian about a threading issue: FB> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=432729 FB> could someone take a look (waffel ? other?) at the submitted patch and FB> commit it if it is ok before the official release. FB> namespace log4cpp { FB> - Appender::AppenderMap* Appender::_allAppenders; FB> threading::Mutex Appender::_appenderMapMutex; FB> /* assume _appenderMapMutex locked */ FB> Appender::AppenderMap& Appender::_getAllAppenders() { FB> - if (!_allAppenders) FB> - _allAppenders = new Appender::AppenderMap(); FB> - FB> + static Appender::AppenderMap* _allAppenders = new FB> Appender::AppenderMap(); FB> return *_allAppenders; FB> } I don't see why folks thinks that it is threading issue, but I see very, very strange bug with line Appender::AppenderMap* Appender::_allAppenders; How it's works in current release I can't understand. This line should be Appender::AppenderMap* Appender::_allAppenders = 0; I have looked the bug report and don't think it is threading issue too. I think we have some problems with NDC. I will look on it closely on weekend. With best regards, Konstantin Litvinenko. |
From: Konstantin L. <Kon...@ma...> - 2007-08-02 13:36:46
|
Hello, fb...@fr...! You wrote to "Konstantin Litvinenko" <Kon...@ma...> on Fri, 27 Jul 2007 11:56:43 +0200: f> could you take a look to: f> https://spaces.internet2.edu/display/SHIB/Log4Cpp I have looked on this and found that there is no differencies between log4cpp CVS HEAD and shibboleth.log4cpp. I have commited bug fix for static initialization into head. Can you please check is CVS HEAD version? There only one thing I can tell is NDC usage in mutitheaded environment is NOT safe. To make it safe user must call log4cpp::getNDC() before any tow threads make simultaneously call to that function. That may be done in main or any other place where is only one thread of exceution that use log4cpp::NDC. With best regards, Konstantin Litvinenko. |