On May 18, 2018 7:20:55 AM EDT, Andrey Rahmatullin <wrar@debian.org> wrote:
On Fri, May 18, 2018 at 01:06:18PM +0200, Andreas Tille wrote:
The reason for this: the configure script compiles the following code:
#include <log4cpp/Category.hh>
#include <iostream>
int main() { std::cout << 1 << '\t' << 0; return 0; }
in order to check that <log4cpp/Category.hh> exists.
But this code still requires -llog4cpp:
Thanks for the explanation but may be I'm missing your point. The
package installs liblog4cpp.a as well and the dynamic library installs
the according .so file. So why should the requriement -llog4cpp not
fulfilled?
The configure script doesn't pass -llog4cpp. It tests for the existence of
the header and tries to find a correct -I option for it. Finding the
correct -L is the next step.
/tmp/cc41MUW4.o: In function `__static_initialization_and_destruction_0(int, int)':
2.cpp:(.text+0x5b): undefined reference to `log4cpp::Appender::AppenderMapStorageInitializer::AppenderMapStorageInitializer()'
2.cpp:(.text+0x70): undefined reference to `log4cpp::Appender::AppenderMapStorageInitializer::~AppenderMapStorageInitializer()'
collect2: error: ld returned 1 exit status
Isn't this rather a bug in log4cpp?
No.