I know this is something so simple I'm going to hate myself for having to ask it, but my head is aching from repeated hits on the desktop. I've read dozens of stackoverflow and google results which suggest that the following should work:
$ ls /usr/local/lib/libusb*
/usr/local/lib/libusb-1.0.a /usr/local/lib/libusb-1.0.so /usr/local/lib/libusb-1.0.so.0.1.0
/usr/local/lib/libusb-1.0.la /usr/local/lib/libusb-1.0.so.0
$ gcc -I ~/libusb-1.0.18/libusb -c test.c
$ gcc -L/usr/local/lib -o test test.o -llibusb
/usr/bin/ld: cannot find -llibusb
collect2: error: ld returned 1 exit status
$ gcc -L/usr/local/lib -o test test.o -llibusb-1.0
/usr/bin/ld: cannot find -llibusb-1.0
collect2: error: ld returned 1 exit status
Why is that not correct? One of those should have worked and I've tried many, many more variations.
For completeness I'm running Ubuntu 14.04 (fresh VM installation).
I built libusb from source (~/libusb-1.0.18) with:
./configure --disable-udev
make
sudo make install
The leading lib and trailing .so are automatically filled in by the linker, so you should not specify either on the command line. Your command should be:
gcc -L/usr/local/lib -o test test.o -lusb-1.0
Related
I have the problem about linking libraries.
"見つかりません" means "Not found"
edit LD_LIBRARY_PATH
$ LD_LIBRARY_PATH="/home/myname/.local/boost/lib" g++ main.cc -lboost_serialization
/home/myname/.linuxbrew/bin/ld: -lboost_serialization が見つかりません
collect2: error: ld returned 1 exit status
edit LIBRARY_PATH
$ LIBRARY_PATH="/home/myname/.local/boost/lib" g++ main.cc -lboost_serialization
/home/myname/.linuxbrew/bin/ld: -lboost_serialization が見つかりません
collect2: error: ld returned 1 exit status
adding the -L option (successfully compiled)
$ g++ main.cc -lboost_serialization -L "/home/myname/.local/boost/lib"
$
Indeed I have successfully compiled my program with the -L option.
But I want to compile without this (like 1 or 2, not 3).
In this case, what do I have to check or modify in my environmental variables?
note: my gcc is 5.3.0 (Homebrew gcc 5.3.0).
I'm trying to make a relocatable object file with gcc. I use solution from this post. The solution works fine with ld:
$ ld -r a.o b.o -o c.o
However when I try to use it with gcc, the following error happens:
$ gcc -r a.o b.o -o c.o
/usr/bin/ld: cannot find -lgcc_s
/usr/bin/ld: cannot find -lgcc_s
collect2: ld returned 1 exit status
Using the -Wl,-r and -Wl,--relocatable options gives the same result.
Is there any way to link relocatable object file with gcc or I'm forced to use ld for doing this?
To solve this problem, the -nostdlib option must also be passed to gcc:
$ gcc -r -nostdlib a.o b.o -o c.o
I don't know it for sure, but it seems without this option gcc tries to link standard libraries into output relocatable object.
I am trying to do something basic, but I can't find the relevant information on how to compile. I tried the following without success:
gfortran testintegral.f90 -lgsl -lgslcblas
testintegral.f90:19.6:
use fgsl
1
Fatal Error: Can't open module file 'fgsl.mod' for reading at (1): No such file
The file is taken from http://de.wikibooks.org/wiki/Fortran:_FGSL#Beispiel:_Numerische_Integration (page in german but readily understandable) so I suppose it is OK.
Maybe the syntax of the compilation command is incorrect ?
EDIT:
I edit my initial post so as not to bury important information in the comments.
Those are the paths of the libraries:
sudo find -name '*libgsl.so*'
./usr/lib/libgsl.so.0
./usr/lib/libgsl.so.0.17.0
sudo find -name '*libgslcblas.so*'
./usr/lib/libgslcblas.so.0
./usr/lib/libgslcblas.so.0.0.0
But I still got an error message when doing:
gfortran testintegral.f90 -L/usr/lib -I/usr/include/fgsl -lfgsl -lgsl -lgslcblas
/usr/bin/ld: cannot find -lgsl
/usr/bin/ld: cannot find -lgslcblas
collect2: error: ld returned 1 exit status
Use the -I flag. For example,
gfortran -I/usr/local/fgsl/include testintegral.f90 -lgsl -lgslcblas
All the .mod files in that directory are then included.
EDIT: See also comments below.
Compilation of a file containing modules in gfortran produces two file types: The source file foo.f90 is translated into foo.o. If foo.f90 contains the modules bar and baz, then bar.mod and baz.mod are also generated. They contain the interface information for these modules. Note that there is no required mapping between module and file names (although programming guildelines may require this).
When the statement use fsgl is found, the interface information is read from fsgl.mod. If that file is not found, you get the error message
Can't open module file 'fgsl.mod' for reading at (1): No such file
So, you have to change your order of compilation (possibly through changing a Makefile).
1) the easiest way is
gfortran testintegral.f90 -I/usr/local/include/fgsl -lfgsl
2) this also works
gfortran -I/usr/local/include/fgsl testintegral.f90 -lgsl -lgslcblas -lm
3) I read the log of the make check in the package, the developer used such a way
gfortran -I/usr/local/include/fgsl -g -O2 -c -o test.o testintegral.f90
/bin/bash /path/.../fgsl-1.3.0/libtool --tag=FC --mode=link gfortran -g -O2 -o test test.o /usr/local/lib/libfgsl.la -lgsl -lgslcblas -lm
UPDATE:
First check the linkers for fgsl
pkg-config --libs fgsl
probably will get something like this
-L/usr/local/lib -lfgsl -lgsl -lgslcblas -lm
Then you put the linkers, works for all the cases!
gfortran -I/usr/include/fgsl example.f90 -lfgsl -lgsl -lgslcblas -lm
UPDATE: I answered too soon, here is the best universal method I found:
gfortran `pkg-config --cflags fgsl` testintegral.f90 -o integral `pkg-config --libs fgsl`
I am trying to compile a sample using gcc under MinGW in windows 7
Why does this work:
$ gcc -m32 -o cube.exe cube.o shader.o matrix.o window.o
/c/dev/mixed/SDKs/Extracted/OpenGLESEmulatorv1.3.0/examples/OpenGLES_20/cube/libEGL.lib
But this doesn't:
$ gcc -m32 -o cube.exe cube.o shader.o matrix.o window.o
-L/c/dev/mixed/SDKs/Extracted/OpenGLESEmulatorv1.3.0/examples/OpenGLES_20/cube
-llibEGL.lib
It fails with:
c:/mingw/bin/../lib/gcc/mingw32/4.7.0/../../../../mingw32/bin/ld.exe: cannot find -llibEGL.lib
collect2.exe: error: ld returned 1 exit status
Shouldn't the -L add the correct search path?
Per the MinGW documentation the -l argument adds lib to the front and .a to the end OR just adds .lib to the end. Removing the .lib from the end allows this to compile.
I'm trying to compile a program using a third party library, Omnet++ in my case. Apparently "make" does not find a library, but the path it uses is correct as you can see (in the sense that I can see the library under omnet++ source tree)
pv135168:basic Bob$ opp_makemake
Creating Makefile in /Users/Bob/Code/network_sim/basic... Makefile created, running "make depend" to add dependencies... opp_makedep -Y --objdirtree -I. -f Makefile -P\$O/ -- ./*.cc
pv135168:basic Bob$ make
g++ -c -g -Wall
-fno-stack-protector -m32 -DHAVE_PCAP -DXMLPARSER=libxml
-DWITH_PARSIM -DWITH_NETBUILDER -I.
-I/Users/Bob/Code/omnetpp-4.1/include -o out/gcc-debug//txc1.o txc1.cc g++ -m32 -Wl,-rpath,/Users/Bob/Code/omnetpp-4.1/lib -Wl,-rpath,. -o out/gcc-debug//basic out/gcc-debug//txc1.o -Wl,-all_load
-L"/Users/Bob/Code/omnetpp-4.1/lib/gcc"
-L"/Users/Bob/Code/omnetpp-4.1/lib" -u _tkenv_lib -lopptkenvd
-loppenvird -lopplayoutd -u _cmdenv_lib -loppcmdenvd -loppenvird
-loppsimd -lstdc++
ld: library not found for -lopptkenvd
collect2: ld returned 1 exit status make: *** [out/gcc-debug//basic]
Error 1 pv135168:basic Bob$
It's looking in the following directories for a file called libopptkenvd.dylib or libopptkenvd.a:
/Users/Bob/Code/omnetpp-4.1/lib/gcc
/Users/Bob/Code/omnetpp-4.1/lib
Is that file in one of those directories (or in the standard directories like /usr/lib)? I don't see an indication of that in your output.