On May 4, 2016 9:32 PM, "Bruce Dubbs" <[email protected]> wrote: > > Douglas R. Reno wrote: >> >> On Wed, May 4, 2016 at 6:16 PM, Douglas R. Reno <[email protected]> >> wrote: >> >>> >>> >>> On Wed, May 4, 2016 at 5:02 PM, Bruce Dubbs <[email protected]> wrote: >>> >>>> Pierre Labastie wrote: >>>> >>>> >>>>> I confirm that GCC-6 is not easy to work with: OpenJDK would not build >>>>> with >>>>> it. I do not have CLANG and have no motivation to build it. Furthermore, >>>>> for >>>>> OpenJDK, the variables CC and CXX are not honoured! What I have done (as >>>>> root): >>>>> >>>>> mv /usr/bin/gcc{,.orig} >>>>> cat >/usr/bin/gcc << EOF >>>>> #!/bin/bash >>>>> gcc -std=c99 "\$@" >>>>> EOF >>>>> mv /usr/bin/g++{,.orig} >>>>> cat >/usr/bin/g++ << EOF >>>>> #!/bin/bash >>>>> gcc -std=c99 "\$@" >>>>> EOF >>>>> >>>>> But that's not enough. The compilation goes to completion, but all the >>>>> compiled executables segfault >>>>> >>>>> Downgrading to gcc-5.3 allowed to build. Tests are currently running, >>>>> but at >>>>> least, the executables seem to run. >>>>> >>>> >>>> I agree that gcc6 will cause us problems. I needed to rebuild ptlib and >>>> ran into errors due to: >>>> >>>> 1. Ambiguous overloaded function. I was able to overcome that by >>>> casting an argument to std::string in several places. Fortunately it >>>> is only one file. src/ptlib/unix/svcproc.cxx >>>> >>>> 2. There is a check for a header that depends on compiler version. It >>>> fails if the minor version is less than 2. That was fixed by adding >>>> || __GNUC__ >= 5 to the #if statement. include/ptlib/critsec.h >>>> >>>> There are also a TON of warnings about deprecated constructs that creates >>>> very noisy output. I was able to turn them off my adding >>>> >>>> STDCCFLAGS += -Wno-deprecated-declarations >>>> >>>> to make/unix.mak. >>>> >>>> It will be easy enough to add these fixes to the existing patch, but I >>>> fear that this will become the rule for a while. >>>> >>>> I'll even note that there are warnings about indentation! >>>> >>>> -- Bruce >>>> >>>> The latest breakage: >>> >>> >>> gst-plugins-good-1.8.1 fails to install/link with taglib-1.11 installed!: >>> >>> make[2]: Entering directory >>> '/sources/gst-plugins-good-1.8.1/gst-plugins-good-1.8.1/ext/taglib' >>> CXXLD libgsttaglib.la >>> /usr/bin/ld: >>> /usr/lib/gcc/x86_64-pc-linux-gnu/6.1.0/../../../../lib64/libtag.a(id3v2framefactory.cpp.o): >>> reolocation R_X86_64_32S against `_ZTVN6TagLib5ID3v212FrameFactoryE' can >>> not be used when making a shared object; recompile with -fPIC >>> /usr/lib/gcc/x86_64-pc-linux-gnu/6.1.0/../../../../lib64/libtag.a: error >>> adding symbols: Bad value >>> collect2: error: ld returned 1 exit status >>> make[2]: *** [libgsttaglib.la] Error 1 >>> make[2]: Leaving directory >>> '/sources/gst-plugins-good-1.8.1/gst-plugins-good-1.8.1/ext/taglib' >>> Makefile:666: recipe for target 'install-recursive' failed >>> make[1]: *** [install-recursive] Error 1 >>> >>> The question is... taglib problem or gst-plugins-good problem? I will do >>> some more research in a little while. >>> >>> I figured it out! >> >> >> The instructions that we have in both books don't compile shared libraries, >> only static ones. The gstreamer build process, at least on my 3 machines, >> won't link to that. In order to fix that, we must add >> -DBUILD_SHARED_LIBS=ON to the cmake command. That seemed to fix it for me. >> >> Am I clear to put in the required changes for this and Qt4 tomorrow? I >> would rather ask first before I commit them, due to prior experiences. > > > I confirm that taglib needs -DBUILD_SHARED_LIBS=ON. Go ahead and add it. > > > -- Bruce > Alright, I will have that done tomorrow afternoon. Part of me wants to try and get Qt4 to compile without Clang still, but C++ is not my specialty. I only know the basics, and Qt is a very complicated suite of programs. It is worth noting that I still can't trace why the va_copy string is becoming undefined, because as far as I can tell, the va_copy call is implemented properly after the va_list call, so if anyone else wants to take a crack at it, I would appreciate seeing what would have to be changed. In the ticket I created on Sunday, I have a link to a log that I put in my home directory on higgs.
-- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
