Compiling on different plattforms.

[Former user]

Hello,

As by change I have been building this software with MSYS2 - 8.3.0 compiler in MINGW64. There are some small issues with cmake and building the library (dll), however this works.

Now I tried to compile with Ubuntu 18.04, I tried both gcc-7 and gcc-8 and both variants. Here I get some issue when building "UncompressedFile.cpp" (Sorry don't have the printout available right now). I can paste it later today.

One question to you, what is your version of compiler and distribution?

Best regards Petter

[Petter Strandh]

I found the reason. Currently vector_BLF only compiles on 64 bit, I know it says so compiling for win 64 bit. Just tried with 32bit and compared one gets these compiler errors.

/vector_blf/src/Vector/BLF/UncompressedFile.cpp: In constructor 'Vector::BLF::UncompressedFile::UncompressedFile()':
/vector_blf/src/Vector/BLF/UncompressedFile.cpp:44:38: warning: overflow in implicit constant conversion [-Woverflow]
m_defaultLogContainerSize(0x20000)

These first lines are not the compile error.

C:/msys64/mingw32/include/c++/7.4.0/bits/postypes.h:178:7: note: candidate: std::fpos<_StateT> std::fpos<_StateT>::operator+(std::streamoff) const [with _StateT = int; std::streamoff = long long int]
operator+(streamoff __off) const
/vector_blf/src/Vector/BLF/UncompressedFile.cpp:73:18: error: ambiguous overload for 'operator+' (operand types are 'std::streampos {aka std::fpos<int>}' and 'const streamsize {aka const int}')
(m_tellg + n > m_fileSize);

‌

It is only UncompressedFile.cpp that is affected by this.

Best regards

Petter

[Tobias Lorenz]

Hello,

As by change I have been building this software with MSYS2 - 8.3.0 compiler in MINGW64. There are some small issues with cmake and building the library (dll), however this works.

Now I tried to compile with Ubuntu 18.04, I tried both gcc-7 and gcc-8 and both variants. Here I get some issue when building "UncompressedFile.cpp" (Sorry don't have the printout available right now). I can paste it later today.

One question to you, what is your version of compiler and distribution?

Best regards Petter

[Tobias Lorenz]

Hi,

I added an integer literal, which declares the 0x20000 in question as being unsigned long int (32-bit).

The change is here: https://bitbucket.org/tobylorenz/vector_blf/branch/explicit_const_type
Can you check if this solves your issue?

Bye
Tobias

[Petter Strandh]

Hello.

No that does not fix the problem. seems to be some thing wrong with something else.

Now I managed to build and I get the same result on Ubuntu 18.04 with multilib and zlib1g-dev:i386, g++-multilib, gcc-multilib

‌

---

‌

Building with:

cmake -DCMAKE_TOOLCHAIN_FILE=../32BIT_TOOLCHAIN.cmake ..

generates: https://pastebin.com/qjMRnB6H

compiling in 64bit generates No errors. Compiling on a 32bit machine. same errors on gcc5, gcc7, gcc8. I am a bit puzzled myself.

File: 32BIT_TOOLCHAIN.cmake

set(CMAKE_SYSTEM_NAME Linux)
set(CMAKE_SYSTEM_VERSION 1)
set(CMAKE_SYSTEM_PROCESSOR "i686")

set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -m32" CACHE STRING "c++ flags")
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -m32" CACHE STRING "c flags")

[Tobias Lorenz]

Hi Petter, issue #15 seems to be related. I provided a fix https://bitbucket.org/tobylorenz/vector_blf/commits/91a0ae0c43332718e9fac467e35dea827c6ce49e that should solve the issue with the operators. Can you check whether this works? Thanks, Tobias

[Petter Strandh]

Hello Tobias,

I can confirm that it builds on MSYS2 32bit mode and on 32bit ARM (arm-linux-gnueabihf) . I also tried the parser example and it does its trick on one of my blf files.

Best regards

Petter

[Tobias Lorenz]

Great. Then I’ll close this ticket.