GCC 4.3.4 and CLANG 3.5 compatiblity - gcc

I am building project which based on C++11 so it needed at least GCC 4.7.2 which is not installed on my platform (and I dont have permission to upgrade software on it). So another option is to build the project with Clang and I downloaded and build CLang 3.5. On platform GCC 4.3.4 is preinstalled.
When I compiling the code I got a lot of errors like that one:
[ 88%] Building CXX object system/CMakeFiles/Grappa.dir/tasks/TaskingScheduler.cpp.o
cd /home/frolo/grappa/build/Make+Release/system && /home/frolo/llvm/build/Debug+Asserts/bin/clang++ -DENABLE_RDMA_AGGREGATOR -DGASNET_CONDUIT_IBV -DGASNET_IBV -DGASNET_SEQ -DSHMMAX=1844674407
3709551615 -std=c++11 -Winline -Wno-inline -mno-red-zone -O3 -g -I/home/frolo/grappa/build/Make+Release/third-party/include -I/home/frolo/grappa/build/Make+Release/third-party/include/ibv-condu
it -I/usr/mpi/gcc/openmpi-1.6.4/include -I/home/frolo/grappa/system -I/home/frolo/grappa/system/tasks -o CMakeFiles/Grappa.dir/tasks/TaskingScheduler.cpp.o -c /home/frolo/grappa/system/tasks
/TaskingScheduler.cpp
In file included from /home/frolo/grappa/system/Barrier.cpp:24:
In file included from /home/frolo/grappa/system/Barrier.hpp:26:
In file included from /home/frolo/grappa/system/ConditionVariable.hpp:27:
In file included from /home/frolo/grappa/system/Message.hpp:27:
In file included from /home/frolo/grappa/system/MessageBase.hpp:33:
In file included from /home/frolo/grappa/system/ConditionVariableLocal.hpp:27:
/home/frolo/grappa/system/tasks/TaskingScheduler.hpp:178:25: error: no matching constructor for initialization of 'std::ofstream' (aka 'basic_ofstream<char>')
std::ofstream f(FLAGS_stats_blob_filename);
^ ~~~~~~~~~~~~~~~~~~~~~~~~~
/usr/lib64/gcc/x86_64-suse-linux/4.3/../../../../include/c++/4.3/fstream:571:7: note: candidate constructor not viable: no known conversion from '::fLS::clstring' (aka 'basic_string<char>') to
'const char *' for 1st argument
basic_ofstream(const char* __s,
^
/usr/lib64/gcc/x86_64-suse-linux/4.3/../../../../include/c++/4.3/bits/fstream.tcc:916:25: note: candidate constructor (the implicit copy constructor) not viable: no known conversion from '::fLS
::clstring' (aka 'basic_string<char>') to 'const std::basic_ofstream<char>' for 1st argument
extern template class basic_ofstream<char>;
^
/usr/lib64/gcc/x86_64-suse-linux/4.3/../../../../include/c++/4.3/fstream:556:7: note: candidate constructor not viable: requires 0 arguments, but 1 was provided
basic_ofstream(): __ostream_type(), _M_filebuf()
^
In file included from /home/frolo/grappa/system/IncoherentAcquirer.cpp:24:
In file included from /home/frolo/grappa/system/IncoherentAcquirer.hpp:28:
In file included from /home/frolo/grappa/system/Message.hpp:27:
In file included from /home/frolo/grappa/system/MessageBase.hpp:33:
In file included from /home/frolo/grappa/system/ConditionVariableLocal.hpp:27:
/home/frolo/grappa/system/tasks/TaskingScheduler.hpp:178:25: error: no matching constructor for initialization of 'std::ofstream' (aka 'basic_ofstream<char>')
std::ofstream f(FLAGS_stats_blob_filename);
^ ~~~~~~~~~~~~~~~~~~~~~~~~~
/usr/lib64/gcc/x86_64-suse-linux/4.3/../../../../include/c++/4.3/fstream:571:7: note: candidate constructor not viable: no known conversion from '::fLS::clstring' (aka 'basic_string<char>') to
'const char *' for 1st argument
basic_ofstream(const char* __s,
^
/usr/lib64/gcc/x86_64-suse-linux/4.3/../../../../include/c++/4.3/bits/fstream.tcc:916:25: note: candidate constructor (the implicit copy constructor) not viable: no known conversion from '::fLS
::clstring' (aka 'basic_string<char>') to 'const std::basic_ofstream<char>' for 1st argument
extern template class basic_ofstream<char>;
...
It looks like incompatibility between gcc 4.3.4 and clang 3.5, does not it? I am not an experienced user of clang, what can be done here to mend it?
Thank you for any help!
Best,
Alex

Clang 3.5 uses the gcc 4.7 ABI so you should compile with gcc >= 4.7. Alternatively, you can download nightly LLVM/Clang builds in .deb format for Ubuntu/Debian based Linux distros from this site.

Related

Can I safely ignore these pthread warnings via glibc?

I am getting a pair of warnings with version 2.22 of glibc:
In file included from /net/module/sw/glibc/2.22/include/pthread.h:23:0,
from foo.h:48,
from foo.c:23:
/net/module/sw/glibc/2.22/include/sched.h:74:57: warning: ‘struct timespec’ declared inside parameter list
extern int sched_rr_get_interval (__pid_t __pid, struct timespec *__t) __THROW;
^
/net/module/sw/glibc/2.22/include/sched.h:74:57: warning: its scope is only this definition or declaration, which is probably not what you want
In file included from foo.h:48:0,
from foo.c:23:
/net/module/sw/glibc/2.22/include/pthread.h:1002:21: warning: ‘struct timespec’ declared inside parameter list
const struct timespec *__restrict __abstime)
^
And:
In file included from /net/module/sw/glibc/2.22/include/sys/param.h:26:0,
from foo.h:51,
from foo.c:23:
/net/module/sw/glibc/2.22/include/limits.h:123:3: warning: #include_next is a GCC extension
# include_next <limits.h>
^
I am using GCC 5.3.0 with this version of glibc.
Since these warnings are referencing an external library I do not control, can I safely ignore these warnings if my application otherwise appears to pass tests?
My concern is that these warnings (particularly those related to pthread) may indicate the introduction of subtle bugs that I do not have tests to catch.
Yes, these warnings should be harmless, but the struct timespec warning is certainly odd. Usually, they would be masked by GCC's warning suppression for system headers.

CppuTest to build with C++11 versions

Installed g++ 4.9.0 (experimental) version under Ubuntu (I am using certain features provided by this version)
When building my code, I use cmake from a script, and it builds correctly. gcc below contains the path where g++ is installed (/mnt...)
$_cmake_ $_dir_ -DCMAKE_CXX_COMPILER=${_gpp_} -DCMAKE_C_COMPILER=${_gcc_} -DCMAKE_BUILD_TYPE=${_build_} -DCMAKE_INSTALL_PREFIX=${_install_} -DBUILD_TESTING:BOOL=1
Now, when I make the CppuTest to build the Unit Tests for my code, I get the following error.
make all
compiling UncrosserTest.cpp
In file included from /home/miguel/Desktop/Ugur/scmProject/scm/dist/dev/mfx_prod/scm/20160207/include/scm/services/primitives.hpp:4:0,
from /home/miguel/Desktop/Ugur/scmProject/log/lidya/src/cpp/scm/services/examples/dummy_strategy/Inside.hpp:6,
from AllTests/Uncrosser/UncrosserTest.cpp:7:
/usr/include/c++/4.9/experimental/optional: In member function ‘void std::experimental::_Optional_base<_Tp, _ShouldProvideDestructor>::_M_construct(_Args&& ...)’:
/usr/include/c++/4.9/experimental/optional:294:18: error: expected type-specifier
::new (std::__addressof(this->_M_payload))
^
/usr/include/c++/4.9/experimental/optional:294:18: error: expected ‘)’
/usr/include/c++/4.9/experimental/optional: In member function ‘void std::experimental::_Optional_base<_Tp, false>::_M_construct(_Args&& ...)’:
/usr/include/c++/4.9/experimental/optional:424:18: error: expected type-specifier
::new (std::__addressof(this->_M_payload))
^
/usr/include/c++/4.9/experimental/optional:424:18: error: expected ‘)’
make: *** [objs/AllTests/Uncrosser/UncrosserTest.o] Error 1
In the CppuTest makefile have included CPPUTEST_CPPFLAGS += -std=c++1y. I guess that am missing some flags in this makefile because as I said the C++ experimental library is building correctly in my "regular" code. What am I missing ?
Possible it's an include problem, see CppUTest Manual - section Conflicts with operator new macros (STL!) and Conflicts with my own overload!
TL;DR
In UncrosserTest add includes for CppUTest after your's.
Instead of
#include <CppUTest/TestHarnes.h>
#include "whatever.h"
do
#include "whatever.h"
#include <CppUTest/TestHarnes.h>

How to make librtmp compile on Mac OS X

I am trying to build rtmpdump and librtmp on mac, however I cannot for the life of me work out why it isn't working. I am using my own built openssl:
$ openssl version
OpenSSL 1.1.0-pre3 (alpha) 15 Feb 2016
I have set the flags correctly (I think), but yet when I try and compile I get this:
$ make INC=-I/usr/local/include
gcc -Wall -I/usr/local/include -DRTMPDUMP_VERSION=\"v2.4\" -DUSE_OPENSSL -O2 -fPIC -c -o rtmp.o rtmp.c
In file included from rtmp.c:152:
./handshake.h:120:12: error: variable has incomplete type 'HMAC_CTX' (aka 'struct hmac_ctx_st')
HMAC_CTX ctx;
^
/usr/local/include/openssl/ossl_typ.h:145:16: note: forward declaration of 'struct hmac_ctx_st'
typedef struct hmac_ctx_st HMAC_CTX;
^
In file included from rtmp.c:152:
./handshake.h:125:3: warning: implicit declaration of function 'HMAC_CTX_init' is invalid in C99 [-Wimplicit-function-declaration]
HMAC_setup(ctx, secretKey, 128);
^
./handshake.h:72:35: note: expanded from macro 'HMAC_setup'
#define HMAC_setup(ctx, key, len) HMAC_CTX_init(&ctx); HMAC_Init_ex(&ctx, key, len, EVP_sha256(), 0)
^
./handshake.h:127:3: warning: implicit declaration of function 'HMAC_CTX_cleanup' is invalid in C99 [-Wimplicit-function-declaration]
HMAC_finish(ctx, digest, digestLen);
^
./handshake.h:74:67: note: expanded from macro 'HMAC_finish'
#define HMAC_finish(ctx, dig, dlen) HMAC_Final(&ctx, dig, &dlen); HMAC_CTX_cleanup(&ctx)
^
./handshake.h:269:12: error: variable has incomplete type 'HMAC_CTX' (aka 'struct hmac_ctx_st')
HMAC_CTX ctx;
^
/usr/local/include/openssl/ossl_typ.h:145:16: note: forward declaration of 'struct hmac_ctx_st'
typedef struct hmac_ctx_st HMAC_CTX;
^
rtmp.c:192:13: warning: comparison of constant 18446744073709551597 with expression of type 'uint32_t' (aka 'unsigned int') is always false [-Wtautological-constant-out-of-range-compare]
if (nSize > SIZE_MAX - RTMP_MAX_HEADER_SIZE)
~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3 warnings and 2 errors generated.
make[1]: *** [rtmp.o] Error 1
make: *** [librtmp/librtmp.a] Error 2
It looks like it might be a compatibility issue with the version of OpenSSL I'm using. Does anyone know how to fix this kind of issue, or any changes to OpenSSL that might be responsible?
OpenSSL 1.1.0-pre3 (alpha) 15 Feb 2016...
Does anyone know how to fix this kind of issue, or any changes to OpenSSL that might be responsible?
Broadly speaking, there appears to be two high level changes in OpenSSL 1.1.x that caused most of the problems. First was the changes to the build system. Second was the firmer posture with respect to private APIs. I think you're seeing an artifact of the second, the private APIs.
Generally speaking, 1.1.0-pre3 and 1.1.0-pre4 have a number of problems and mostly can't achieve a successful make test. The issues include configuration and build issues for 32-bit OS X PowerMac, 64-bit OS X PowerMac, 32-bit OS X, 64-bit OS X, 32-bit ARM, 64-bit ARM, OpenBSD 5.7, NetBSD 7.0, Cygwin, CentOS 5, CentOS 7 and i686. So many so that I've filed about 25 bugs in preparation for the upcoming Beta-1 scheduled to arrive Monday, 14 MAR 2016.
You should probably avoid 1.1.0-pre-N at the moment because its so unstable. You should be OK under x86_64 Linux, but anything else is a crap shoot at the moment. You should be able to pick it up again around Beta-2.
You will probably have better luck with the tip of Master after commits from Friday, 11 MAR 2016. That would be roughly commits 0d4d5ab and 4c1cf7e. You can get the latest with:
git clone git://git.openssl.org/openssl.git
Or, use 1.0.2 stable releases.

Some C++11 features missing when using Code Blocks 13.12, MinGW 4.8.1 and SFML 2.1

Turning to Stack Overflow after exhausting all my efforts.
I'm running Code Blocks 13.12 and MinGW 4.8.1 on a Windows 7 OS 64bit system. I spent two days trying to get the compiler to link (statically) to the SFML 2.1 library. I followed both the SFML tutorial and this video tutorial- reinstalling both Code Blocks and MinGW twice before it finally linked.
After that I was relieved that SFML was recognized, but I got 50 errors when I tried to build and run that suggested C++11 features weren't.
Added -std=gnu++11 to compiler options and 13 of those errors went away.
The remaining errors all fall into two categories: mutex is not a member of std and from time t is not a member of chrono::_V2::steady clock. I'm not even sure what _V2 is.
What am I missing?
-------------- Build: Release in PumpTracker2.0 (compiler: GNU GCC Compiler)---------------
mingw32-g++.exe -Wall -std=gnu++11 -DSFML_STATIC -O2 -std=gnu++11 -IC:\SFML-2.1\include -IC:\SFML-2.1\include -c C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp -o obj\Release\main.o
In file included from C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp:25:0:
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\Account.h:37:1: error: 'mutex' in namespace 'std' does not name a type
std::mutex mu;
^
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\Account.h:86:22: error: 'to_time_t' is not a member of 'std::chrono::_V2::steady_clock'
time_t currentTp = std::chrono::steady_clock::to_time_t(newly);
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\Account.h:94:48: error: 'from_time_t' is not a member of 'std::chrono::_V2::steady_clock'
std::chrono::steady_clock::time_point tNew = std::chrono::steady_clock::from_time_t(currentTp);
^
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp:51:2: error: 'mutex' is not a member of 'std'
std::mutex mu3;
^
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp:51:13: error: expected ';' before 'mu3'
std::mutex mu3;
^
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp:52:19: error: 'mutex' was not declared in this scope
std::unique_lock<mutex> locker3(mu3, std::defer_lock);
^
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp:52:24: error: template argument 1 is invalid
std::unique_lock<mutex> locker3(mu3, std::defer_lock);
^
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp:52:33: error: invalid type in declaration before '(' token
std::unique_lock<mutex> locker3(mu3, std::defer_lock);
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp:214:22: error: 'to_time_t' is not a member of 'std::chrono::_V2::steady_clock'
pS->tpNewest = std::chrono::steady_clock::to_time_t(NextPump);
^
C:\Users\WITcom\Desktop\C++\PumpTracker2.0\main.cpp:245:51: error: 'from_time_t' is not a member of 'std::chrono::_V2::steady_clock'
std::chrono::steady_clock::time_point TT2 = std::chrono::steady_clock::from_time_t(pS->tpNewest);
^
Process terminated with status 1 (0 minute(s), 4 second(s))
37 error(s), 47 warning(s) (0 minute(s), 4 second(s))
As already mentioned the original MinGW distribution is not the best one to go with, if you want to go with up-to-date features. Instead there are projects like Nuwen (Stephan T. Lavavej's website) as well as MinGW Builds that provide MinGW-w64 binaries.
The mentioning of SFML is misleading, since the problems have nothing to do with SFML at all. Keep in mind however that you'll have to rebuild SFML if you change the compiler!
As a next step, it would've been important to post a minimal example of the source code that reproduces the errors, because right now, we can only guess what's going wrong, based on the errors.
If you don't use the GNU extensions explicitly, you should use -std=c++11 instead of -std=gnu++11 and if you look at the build command, you'll notice that it's included twice, which is not necessary.
And now to some guessing about the errors:
Account.h - Make sure the <mutex> header is included.
main.cpp - Make sure the <mutex> header is included.
main.cpp - If std::unique_lock<mutex> is from your code, then you need to add std::, i.e. std::unique_lock<std::mutex>.
main.cpp - time_t things seem to be resolved, but make sure anyways to include <chrono> and <ctime> (since time_t is defined in <ctime>).

ROracle installation fails on R 2.15.1 using Rtools 2.15

I'm trying to compile the ROracle package version 1.1-4 under Windows using R 2.15.1 and Rtools 2.15
install.packages('ROracle',type='source')
but it seems that ocimingw.c is missing (see output below). I get the same error message using R CMD INSTALL --build.
edit 2012-09-11: The problem is apparently specific to 1.1-4, as I managed to build and install 1.1-1 normally.
Any idea what the solution might be?
trying URL 'http://stat.ethz.ch/CRAN/src/contrib/ROracle_1.1-4.tar.gz'
Content type 'application/x-gzip' length 156924 bytes (153 Kb)
opened URL
downloaded 153 Kb
* installing *source* package 'ROracle' ...
** package 'ROracle' successfully unpacked and MD5 sums checked
looking for Instant Client c:/instantclient/x64/instantclient_11_2
cygwin warning:
MS-DOS style path detected: C:\instantclient\x64\instantclient_11_2
Preferred POSIX equivalent is: /cygdrive/c/instantclient/x64/instantclient_11_2
CYGWIN environment variable option "nodosfilewarning" turns off this warning.
Consult the user's guide for more details about POSIX paths:
http://cygwin.com/cygwin-ug-net/using.html#using-pathnames
found Instant Client C:\instantclient\x64\instantclient_11_2
found Instant Client SDK c:/instantclient/x64/instantclient_11_2/sdk/include
generating ocimingw.lib from ocimingw.def
x86_64-w64-mingw32-dlltool: not found
copying from c:/instantclient/x64/instantclient_11_2/sdk/include
** libs
Warning: this package has a non-empty 'configure.win' file,
so building only the main architecture
cygwin warning:
MS-DOS style path detected: C:/R/R-215~1.1/etc/x64/Makeconf
Preferred POSIX equivalent is: /cygdrive/c/R/R-215~1.1/etc/x64/Makeconf
CYGWIN environment variable option "nodosfilewarning" turns off this warning.
Consult the user's guide for more details about POSIX paths:
http://cygwin.com/cygwin-ug-net/using.html#using-pathnames
gcc -m64 -I"C:/R/R-215~1.1/include" -DNDEBUG -I./oci -I"d:/RCompile/CRANpkg/extralibs64/local/include" -O2 -Wall -std=gnu99 -mtune=core2 -c rodbi.c -o rodbi.o
rodbi.c: In function 'rociResExec':
rodbi.c:1169:14: warning: unused variable 'drv' [-Wunused-variable]
rodbi.c: In function 'rociResFetch':
rodbi.c:1187:16: warning: unused variable 'drv' [-Wunused-variable]
rodbi.c: In function 'rociResInfo':
rodbi.c:1256:16: warning: unused variable 'drv' [-Wunused-variable]
rodbi.c: In function 'rociResTerm':
rodbi.c:1323:14: warning: unused variable 'drv' [-Wunused-variable]
rodbi.c: In function 'rodbiCheck':
rodbi.c:2348:25: warning: pointer targets in passing argument 1 of 'snprintf' differ in signedness [-Wpointer-sign]
c:\rtools\gcc-4.6.3\bin\../lib/gcc/i686-w64-mingw32/4.6.3/../../../../i686-w64-mingw32/include/stdio.h:505:15: note: expected 'char * restrict' but argument is of type 'text *'
rodbi.c: In function 'rawToStrhexCall':
rodbi.c:2573:7: warning: pointer targets in passing argument 1 of 'Rf_mkChar' differ in signedness [-Wpointer-sign]
C:/R/R-215~1.1/include/Rinternals.h:628:6: note: expected 'const char *' but argument is of type 'unsigned char *'
rodbi.c:2557:26: warning: unused variable 'seglen' [-Wunused-variable]
rodbi.c: In function 'rociResInit':
rodbi.c:1121:3: warning: 'qry_encoding' may be used uninitialized in this function [-Wuninitialized]
gcc -m64 -I"C:/R/R-215~1.1/include" -DNDEBUG -I./oci -I"d:/RCompile/CRANpkg/extralibs64/local/include" -O2 -Wall -std=gnu99 -mtune=core2 -c rooci.c -o rooci.o
In file included from C:/R/R-215~1.1/include/R.h:44:0,
from rodbi.h:37,
from rooci.c:47:
C:/R/R-215~1.1/include/R_ext/RS.h:43:0: warning: "ERROR" redefined [enabled by default]
c:\rtools\gcc-4.6.3\bin\../lib/gcc/i686-w64-mingw32/4.6.3/../../../../i686-w64-mingw32/include/wingdi.h:70:0: note: this is the location of the previous definition
rooci.c: In function 'roociInitializeRes':
rooci.c:620:32: warning: 'cnvcid' may be used uninitialized in this function [-Wuninitialized]
gcc -m64 -shared -s -static-libgcc -o ROracle.dll tmp.def rodbi.o rooci.o -L. ocimingw.lib -Ld:/RCompile/CRANpkg/extralibs64/local/lib/x64 -Ld:/RCompile/CRANpkg/extralibs64/local/lib -LC:/R/R-215~1.1/bin/x64 -lR
gcc.exe: error: ocimingw.lib: No such file or directory
ERROR: compilation failed for package 'ROracle'
* removing 'C:/R/R-2.15.1/library/ROracle'
The problem is apparently specific to 1.1-4, as I managed to build and install 1.1-1 normally. I will wait for the 1.1-5 version before I give it another try. Thanks to all contributors!
ROracle is quite easy to install from source in Windows, try using Linux if you want to using ROracle from source code, here is my install log.
If still perfer windows, you can get a compiled ROracle package from Oracle, it's included in the "Client Supporting Packages".

Resources