Re: [Qucs-devel] Test simplify_20131114 branch

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 454-5900

On 04/12/13 22:53, Bastien ROUCARIES wrote:
> On Wed, Dec 4, 2013 at 10:32 PM, Guilherme Brondani Torri
> <gui...@gm...> wrote:
>> On 04/12/13 21:47, Bastien ROUCARIES wrote:
>>> Hi,
>>>
>>> Could you test the branch simplify_20131114 ?
>>>
>>> It simplify a lot qucs:
>>> 205 files changed, 4893 insertions(+), 6970 deletions(-)
>>>
>>> So a net deletion of about 2000 loc.
>>>
>>> Bastien
>>>
>> Hi again,
>>
>> Got around the other issue, but I will let you add the ax_cxx_namespaces.m4
>> to qucs-core/m4.
>>
>> Now it got stuck at:
>> <snip>
>> /bin/sh ../../../libtool --tag=CXX --mode=compile g++ -DHAVE_CONFIG_H -I.
>> -I../../.. -I/usr/local/Cellar/eigen/3.2.0/include/eigen3 -I../../../src
>> -I../../../src/math -I../../../src/components
>> -I../../../src/components/devices -g -O2 -pipe -fno-exceptions -fno-rtti
>> -fno-check-new -W -Wall -MT substrate.lo -MD -MP -MF .deps/substrate.Tpo -c
>> -o substrate.lo substrate.cpp
>> libtool: compile: g++ -DHAVE_CONFIG_H -I. -I../../..
>> -I/usr/local/Cellar/eigen/3.2.0/include/eigen3 -I../../../src
>> -I../../../src/math -I../../../src/components
>> -I../../../src/components/devices -g -O2 -pipe -fno-exceptions -fno-rtti
>> -fno-check-new -W -Wall -MT substrate.lo -MD -MP -MF .deps/substrate.Tpo -c
>> substrate.cpp -fno-common -DPIC -o .libs/substrate.o
>> In file included from ../../../src/net.h:28,
>> from ../../../src/components/component.h:42,
>> from substrate.cpp:29:
>> ../../../src/ptrlist.h:31: error: expected unqualified-id before ‘using’
>> In file included from ../../../src/components/component.h:42,
>> from substrate.cpp:29:
>> ../../../src/net.h:75: error: ISO C++ forbids declaration of ‘ptrlist’ with
>> no type
>> ../../../src/net.h:75: error: expected ‘;’ before ‘<’ token
>> ../../../src/net.h:94: error: ISO C++ forbids declaration of ‘ptrlist’ with
>> no type
>> ../../../src/net.h:94: error: expected ‘;’ before ‘<’ token
>> ../../../src/net.h:95: error: ISO C++ forbids declaration of ‘ptrlist’ with
>> no type
>> ../../../src/net.h:95: error: expected ‘;’ before ‘<’ token
>> In file included from ../../../src/characteristic.h:28,
>> from ../../../src/circuit.h:28,
>> from ../../../src/components/component.h:43,
>> from substrate.cpp:29:
>> ../../../src/pair.h: In constructor ‘pair::pair(const char*)’:
>> ../../../src/pair.h:39: error: ‘nullptr’ was not declared in this scope
>> ../../../src/pair.h: In constructor ‘pair::pair(const char*, double)’:
>> ../../../src/pair.h:43: error: ‘nullptr’ was not declared in this scope
>> ../../../src/pair.h: In member function ‘void pair::setName(const char*)’:
>> ../../../src/pair.h:47: error: ‘nullptr’ was not declared in this scope
>> make[4]: *** [substrate.lo] Error 1
>> make[3]: *** [all-recursive] Error 1
>> make[2]: *** [all-recursive] Error 1
>> make[1]: *** [all-recursive] Error 1
>> make: *** [all] Error 2
>>
>> ~/git/qucs/qucs-core $ g++ -v
>> Using built-in specs.
>> Target: i686-apple-darwin11
>> Configured with:
>> /private/var/tmp/llvmgcc42/llvmgcc42-2336.11~182/src/configure
>> --disable-checking --enable-werror
>> --prefix=/Applications/Xcode.app/Contents/Developer/usr/llvm-gcc-4.2
>> --mandir=/share/man --enable-languages=c,objc,c++,obj-c++
>> --program-prefix=llvm- --program-transform-name=/^[cg][^.-]*$/s/$/-4.2/
>> --with-slibdir=/usr/lib --build=i686-apple-darwin11
>> --enable-llvm=/private/var/tmp/llvmgcc42/llvmgcc42-2336.11~182/dst-llvmCore/Developer/usr/local
>> --program-prefix=i686-apple-darwin11- --host=x86_64-apple-darwin11
>> --target=i686-apple-darwin11 --with-gxx-include-dir=/usr/include/c++/4.2.1
>> Thread model: posix
>> gcc version 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.11.00)
>>
>>
>> I didn't read all your changes, but I see C++11 on the log.
>> Just curious, do you know if there is a minimum version of GCC or LLVM Clang
>> for the features you added?
> Nothing really advanced.
>
> Unfortunatly alias template need gcc 4.7 or clang 3.1
>
> This feature is needed in order to alias eigen to tmatrix.
>
> Bastien
>> Regards,
>> Guilherme

Hi Bastien,

Raising the bar on the minimum compiler version is ok with me. I guess 
GCC 4.7 is already on stable Debian, no?

On the Mac we can use Clang, after working out the bugs. I guess the GCC 
4.2 from Apple won't cut it.

On Windows, hum, for native compilation the issue is Qt. The binary 
libraries shipped by Digia require GCC 4.4. Off course one can compile 
from source, but it is faster/easier to use the official binaries... Not 
ideal, but one can also use a newer version just for qucs-core.

What the other developers have to say?

Ok back to your code, trying with Clang it went a bit further. A few 
notes below, till the point where I got stuck again.

Compiler version:
~/git/qucs/qucs-core $ clang++ -v
Apple LLVM version 4.2 (clang-425.0.28) (based on LLVM 3.2svn)
Target: x86_64-apple-darwin12.5.0

0) On the Mac I am using:
$ CXX=clang++ CXXFLAGS="-std=c++11 -stdlib=libc++" 
LDDFLAGS=-stdlib=libc++ ./configure --enable-maintainer-mode

If I don't enforce the CXX, CXXFLAGS and LDDFLAGS, see what happens:

/bin/sh ../../libtool  --tag=CXX   --mode=compile clang++ 
-DHAVE_CONFIG_H -I. -I../.. 
-I/usr/local/Cellar/eigen/3.2.0/include/eigen3  -I../../src   -g -O2 
-std=gnu++11 -pipe -fno-exceptions -fno-rtti -fno-check-new -W -Wall -MT 
matrix.lo -MD -MP -MF .deps/matrix.Tpo -c -o matrix.lo matrix.cpp
libtool: compile:  clang++ -DHAVE_CONFIG_H -I. -I../.. 
-I/usr/local/Cellar/eigen/3.2.0/include/eigen3 -I../../src -g -O2 
-std=gnu++11 -pipe -fno-exceptions -fno-rtti -fno-check-new -W -Wall -MT 
matrix.lo -MD -MP -MF .deps/matrix.Tpo -c matrix.cpp -fno-common -DPIC 
-o .libs/matrix.o
clang: warning: argument unused during compilation: '-fno-check-new'
In file included from matrix.cpp:96:
In file included from ./matrix.h:28:
In file included from 
/usr/local/Cellar/eigen/3.2.0/include/eigen3/Eigen/Dense:1:
In file included from 
/usr/local/Cellar/eigen/3.2.0/include/eigen3/Eigen/Core:256:
/usr/local/Cellar/eigen/3.2.0/include/eigen3/Eigen/src/Core/util/Memory.h:737:23: 
error: no member named 'forward' in namespace 'std'
       ::new(p) T(std::forward<Args>(args)...);
                  ~~~~~^
/usr/local/Cellar/eigen/3.2.0/include/eigen3/Eigen/src/Core/util/Memory.h:737:31: 
error: 'Args' does not refer to a value
       ::new(p) T(std::forward<Args>(args)...);
                               ^
/usr/local/Cellar/eigen/3.2.0/include/eigen3/Eigen/src/Core/util/Memory.h:734:26: 
note: declared here
     template<typename... Args>
                          ^
2 errors generated.

This was easy to Google out.

1) It seems that TR1 is not available/necessary with Clang (or C++11):

Out of the configure phase.

<snip>
checking for EIGEN3... yes
checking whether the compiler implements namespaces... yes
checking whether the compiler has numeric_limits<T>... yes
checking whether the compiler implements namespaces... yes
checking whether the compiler has complex<T>... yes
checking whether the compiler has TR1 compliant complex<T>... no
checking for complex cos... yes
checking for complex cosh... yes
checking for complex exp... yes
checking for complex log... yes
checking for complex log10... yes
checking for complex sin... yes
checking for complex sinh... yes
checking for complex sqrt... yes
checking for complex tan... yes
checking for complex tanh... yes
checking for complex acos... yes
checking for complex acosh... yes
checking for complex asin... yes
checking for complex asinh... yes
checking for complex atan... yes
checking for complex atanh... yes
checking for complex log2... no
checking for complex norm... yes
checking for complex pow... yes
checking for complex atan2... no
checking for complex fmod... no
checking for complex polar (double, double)... yes
checking for complex polar (complex, complex)... no
<snip>

The build moves forward to stop with:

<snip>
/bin/sh ../../libtool  --tag=CXX   --mode=compile clang++ 
-DHAVE_CONFIG_H -I. -I../.. 
-I/usr/local/Cellar/eigen/3.2.0/include/eigen3  -I../../src -std=c++11 
-stdlib=libc++ -pipe -fno-exceptions -fno-rtti -fno-check-new -W -Wall 
-MT complex.lo -MD -MP -MF .deps/complex.Tpo -c -o complex.lo complex.cpp
libtool: compile:  clang++ -DHAVE_CONFIG_H -I. -I../.. 
-I/usr/local/Cellar/eigen/3.2.0/include/eigen3 -I../../src -std=c++11 
-stdlib=libc++ -pipe -fno-exceptions -fno-rtti -fno-check-new -W -Wall 
-MT complex.lo -MD -MP -MF .deps/complex.Tpo -c complex.cpp  -fno-common 
-DPIC -o .libs/complex.o
clang: warning: argument unused during compilation: '-fno-check-new'
In file included from complex.cpp:37:
./complex.h:50:12: error: no member named 'tr1' in namespace 'std'
using std::tr1::acos;
       ~~~~~^
./complex.h:56:12: error: no member named 'tr1' in namespace 'std'
using std::tr1::acosh;
       ~~~~~^
./complex.h:62:12: error: no member named 'tr1' in namespace 'std'
using std::tr1::asin;
       ~~~~~^
./complex.h:68:12: error: no member named 'tr1' in namespace 'std'
using std::tr1::asinh;
       ~~~~~^
./complex.h:74:12: error: no member named 'tr1' in namespace 'std'
using std::tr1::atan;
       ~~~~~^
./complex.h:80:12: error: no member named 'tr1' in namespace 'std'
using std::tr1::atanh;
       ~~~~~^
complex.cpp:333:25: error: no member named 'tr1' in namespace 'std'
   nr_complex_t a = std::tr1::atan (y / x);
                    ~~~~~^
7 errors generated.
<snip>

Looking at the error line in `complex.h`
I don't get this, it should fall into the first branch of`#ifndef 
HAVE_CXX_COMPLEX_ACOS`. 'Acos' is supposed to be defined as seen above 
in the configure.

Anyway, removing the 'tr1::' from the second branch lets it go ahead. It 
seems that the ifdefs on complex.h are not dealing with the fact that 
TRI is now integrated into C++11.

2) Finally it gets stuck at:

<snip>
/bin/sh ../../../libtool  --tag=CXX   --mode=compile clang++ 
-DHAVE_CONFIG_H -I. -I../../.. 
-I/usr/local/Cellar/eigen/3.2.0/include/eigen3  -I../../../src 
-I../../../src/math -I../../../src/components 
-I../../../src/components/devices   -std=c++11 -stdlib=libc++ -pipe 
-fno-exceptions -fno-rtti -fno-check-new -W -Wall -MT bondwire.lo -MD 
-MP -MF .deps/bondwire.Tpo -c -o bondwire.lo bondwire.cpp
libtool: compile:  clang++ -DHAVE_CONFIG_H -I. -I../../.. 
-I/usr/local/Cellar/eigen/3.2.0/include/eigen3 -I../../../src 
-I../../../src/math -I../../../src/components 
-I../../../src/components/devices -std=c++11 -stdlib=libc++ -pipe 
-fno-exceptions -fno-rtti -fno-check-new -W -Wall -MT bondwire.lo -MD 
-MP -MF .deps/bondwire.Tpo -c bondwire.cpp  -fno-common -DPIC -o 
.libs/bondwire.o
clang: warning: argument unused during compilation: '-fno-check-new'
bondwire.cpp:404:47: error: call to 'norm' is ambiguous
   nr_double_t f = kelvin (T) * 4.0 * R * z0 / norm (4.0 * z0 + R) / T0;
                                               ^~~~
../../../src/math/real.h:31:15: note: candidate function
nr_double_t   norm (const nr_double_t);
               ^
/usr/bin/../lib/c++/v1/complex:953:1: note: candidate function
norm(double __re)
^
/usr/bin/../lib/c++/v1/complex:946:1: note: candidate function
norm(long double __re)
^
/usr/bin/../lib/c++/v1/complex:972:1: note: candidate function
norm(float __re)
^
1 error generated.
<snip>

Any idea how to get past here? Do we need to cast these floats? Of there 
is a clever way of doing it?

Regards,
Guilherme

Re: [Qucs-devel] Test simplify_20131114 branch

A circuit simulator with graphical user interface (GUI)

Re: [Qucs-devel] Test simplify_20131114 branch