Thanks a lot for the response again... I will try sun's compiler next...
But for now... Once I used the cvs code, stuff looks different... I have
not seen the errors i got with the dev package, but different ones:
gmake CXXFLAGS=-DMSG_NOSIGNAL=0
g++ -DMSG_NOSIGNAL=0 -Iinclude -fPIC -c -o src/Common.o src/Common.cpp
src/Common.cpp:334: error: duplicate explicit instantiation of
`std::wstring CarobNS::toWString(const T&) [with T = CarobNS::java_byte]'
src/Common.cpp:356: error: duplicate explicit instantiation of
`std::wstring CarobNS::toUserString(const T&) [with T = CarobNS::java_byte]'
gmake: *** [src/Common.o] Error 1
This happens right away as you can see...
In my previous attempt with the non-cvs version I got the whole thing to
compile... But the way I solved the std::signbit issue was obviously
wrong, as the linker complained then of multiple declarations... Yousee,
the way i tried to solve it is by declaring it, which worked for
compilation, but not for linking... Obviously something in the code did
know how to include that.
Anyways, I can't get that far with current cvs checkout...
Thanks again in advance for your help!
Herbert wrote:
Alexander Wallace <[EMAIL PROTECTED]> writes:
So what i did is (not really knowing if this is really what should
happen... Is the following
Added this to include/SystemDependantDefs.hpp
[
const int MSG_NOSIGNAL = 0;
This is a nasty workaround that may break failover capabilities
sending uncaught SIGPIPE signals. Please leave it ON until we find an
appropriate fix, so we can make progress on other fronts meanwhile.
On the latest code try this variant instead:
make CXXFLAGS=-DMSG_NOSIGNAL=0
(not modifying the code)
const int SOCKET_CREATION_DOMAIN = 2;
]
This one should not be needed anymore on the latest CVS code, could
you try without it?
And to src/JavaSocket.cpp
[
#define __LITTLE_ENDIAN 1234
#define __BIG_ENDIAN 4321
#if defined(_BIG_ENDIAN)
#define __BYTE_ORDER __BIG_ENDIAN
#else
#define __BYTE_ORDER __LITTLE_ENDIAN
#endif
#include <sys/int_types.h>
/*
* Some types
*/
typedef uint8_t u_int8_t;
typedef uint16_t u_int16_t;
typedef uint32_t u_int32_t;
typedef uint64_t u_int64_t;
]
Those look fine. We may put them into a solaris.h header.
src/ParameterStatement.cpp:38: error: `std::signbit' has not been declared
src/ParameterStatement.cpp:312: error: `isinf' undeclared (first use
this function)
We are interested to know how you solved these ones.
src/SQLDataSerialization.cpp:39:2: #error "only IEEE 754 platforms are
supported"
gmake: *** [src/SQLDataSerialization.o] Error 1
pretty sad isn't it?
Not so much, this is just a paranoid sanity check that you can/we will
safely remove. Only very old platforms are not IEEE754 compliant
Marc Herbert <[EMAIL PROTECTED]> writes:
Do you also have a Sun C compiler installed? We try to be as standard
conformant as possible and use almost zero gcc-specific trick. So in
case your gcc installation is not as well integrated in your system as
your Sun compiler is (just guessing...), the Sun compiler may even
work better, despite us using mainly gcc.
To try another compiler, run this:
make CXX=suncc CXXFLAGS=
_______________________________________________
Carob mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/carob
_______________________________________________
Carob mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/carob
