Recent changes to bugs

int64_t issues on 64 bit system

Abe Bachrach — Tue, 17 Jan 2012 16:29:23 -0000

Hi,
On my 64 bit system (gcc version 4.6.1 (Ubuntu/Linaro 4.6.1-9ubuntu3)) , int64_t is typedefed to "long int".

As a result, fast format calls this integer_shim:

inline ximpl_fast_integers::integer_shim<12> filter_type(
signed long value
, signed long const*
, ff_char_t const volatile*
)
{
ximpl_fast_integers::integer_shim<12> r;

r.assign(stlsoft::integer_to_string(r.buff, STLSOFT_NUM_ELEMENTS(r.buff), value, r.len));

return r;
}

which is incorrect!

Is there a good reason to NOT to include <stdint.h> and be explicit about the integer type sizes?
ie: use int64_t instead of "signed long"?

header include ordering error with fastformat.hpp

Abe Bachrach — Thu, 22 Dec 2011 19:58:35 -0000

I'm seeing an error if I include fastformat/fastformat.hpp after fastformat/sinks/stringstream.hpp

ie:
this works:
#include <fastformat/sinks/stringstream.hpp>
#include <fastformat/fastformat.hpp>

this doesn't:
#include <fastformat/fastformat.hpp>
#include <fastformat/sinks/stringstream.hpp>

I get the following compilation error

In file included from /tmp/build/include/fastformat/fastformat.hpp:193:0,
from /tmp/fastformat-util/src/test/ff_test.cpp:30:
/tmp/build/include/fastformat/shims/action/fmt_slices/generic_string.hpp: In function ‘S& fastformat::sinks::fmt_slices(S&, int, size_t, size_t, const fastformat::ff_string_slice_t*) [with S = std::basic_stringstream<char>, size_t = long unsigned int]’:
/tmp/build/include/fastformat/internal/generated/helper_functions.hpp:59:9: instantiated from ‘S& fastformat::internal::helpers::fmt_inner_helper_1(S&, int, const fastformat::ff_format_element_t*, size_t, size_t, const ff_char_t*, size_t) [with S = std::basic_stringstream<char>, size_t = long unsigned int, fastformat::ff_char_t = char]’
/tmp/build/include/fastformat/internal/generated/helper_functions.hpp:70:38: instantiated from ‘S& fastformat::internal::helpers::fmt_outer_helper_1(S&, int, const fastformat::ff_format_element_t*, size_t, size_t, const A0&) [with S = std::basic_stringstream<char>, A0 = char [5], size_t = long unsigned int]’
/tmp/build/include/fastformat/internal/generated/api_functions.hpp:111:90: instantiated from ‘S& fastformat::fmtln(S&, const F&, const A0&) [with S = std::basic_stringstream<char>, F = char [33], A0 = char [5]]’
/tmp/fastformat-util/src/test/ff_test.cpp:56:59: instantiated from here
/tmp/build/include/fastformat/shims/action/fmt_slices/generic_string.hpp:118:5: error: ‘struct std::basic_stringstream<char>’ has no member named ‘reserve’
/tmp/build/include/fastformat/shims/action/fmt_slices/generic_string.hpp:118:5: error: ‘struct std::basic_stringstream<char>’ has no member named ‘size’
/tmp/build/include/fastformat/shims/action/fmt_slices/generic_string.hpp:126:13: error: ‘struct std::basic_stringstream<char>’ has no member named ‘append’
/tmp/build/include/fastformat/shims/action/fmt_slices/generic_string.hpp:134:9: error: ‘struct std::basic_stringstream<char>’ has no member named ‘append’
make[3]: *** [src/test/CMakeFiles/fastformat-test.dir/ff_test.cpp.o] Error 1
make[2]: *** [src/test/CMakeFiles/fastformat-test.dir/all] Error 2
make[1]: *** [all] Error 2
make: *** [all] Error 2

Build fails in Debian wheezy (64 bits) with gcc 4.6.2

Zara — Fri, 09 Dec 2011 22:21:22 -0000

When compiling with 'make build', we get the following error:

g++ -c -Wall -pedantic -Wno-long-long -Wno-unused-value -Wno-non-virtual-dtor -Werror -D_REENTRANT -DNDEBUG -U_DEBUG -O3 -Dunix -DFASTFORMAT_NO_IMPLICIT_LINK -I../../include -I/home/zara/curro/stlsoft/include -o test.performance.scenario1.mt.o ../../test/performance/test.performance.overload.ex.1/test.performance.overload.ex.1.cpp
/home/zara/curro/stlsoft/include/stlsoft/conversion/integer_to_string.hpp: In function ‘int main_(int, char**)’:
/home/zara/curro/stlsoft/include/stlsoft/conversion/integer_to_string.hpp:359:13: error: array subscript is below array bounds [-Werror=array-bounds]
/home/zara/curro/stlsoft/include/stlsoft/conversion/integer_to_string.hpp:359:13: error: array subscript is below array bounds [-Werror=array-bounds]
cc1plus: all warnings being treated as errors

It can be turned off, allowing the build to complete without further errors with the following change:

--- a/build/gcc46.unix/makefile
+++ b/build/gcc46.unix/makefile
@@ -140,7 +140,7 @@ COMMON_INCLUDES =

CPPC_F_NOLOGO =
CPPC_F_COMPILE_ONLY = -c
-CPPC_F_WARNING_MAX = -Wall -pedantic -Wno-long-long -Wno-unused-value -Wno-non-virtual-dtor
+CPPC_F_WARNING_MAX = -Wall -pedantic -Wno-long-long -Wno-unused-value -Wno-non-virtual-dtor -Wno-array-bounds
CPPC_F_WARNING_ERR = -Werror
CPPC_F_EXCEPTIONS_ON =
CPPC_F_EXCEPTIONS_OFF =

It is possibly too harsh, but it works. If you need me to make any further testing, please do tell me, I'll help with pleasure.

Thanks for all your good work!

weird exception behaviour

Anonymous — Thu, 01 Dec 2011 20:52:04 -0000

Tests that catch expected exceptions are failing with GCC 4.2 on Mac OS-X 10.7. Seems to be a code-generation error.

The following changes to xTests' main header work:

# if defined(STLSOFT_COMPILER_IS_GCC)
# define XTESTS_INVOKE_TEST_CASE_FN_INNER_(fn) do { try { (*fn)(); } catch(...) { throw; } } while(0)
# else /* ? compiler */
# define XTESTS_INVOKE_TEST_CASE_FN_INNER_(fn) (*fn)()
# endif /* compiler */

# define XTESTS_INVOKE_TEST_CASE_FN_(fn, name) \ \ do \ { \ try \ { \ XTESTS_INVOKE_TEST_CASE_FN_INNER_(fn); \ } \ catch(std::bad_alloc& x) \ { \ XTESTS_REPORT_EXCEPTION_(x); \ \ throw; \ } \ catch(XTESTS_NS_CPP_QUAL(requirement_failed_exception)& /* x */) \ { \ } \ catch(std::exception& x) \ { \ XTESTS_REPORT_EXCEPTION_(x); \

} \ } while(0)

# define XTESTS_INVOKE_TEST_CASE_FN_THROWS_(fn, name, type) \ \ do \ { \ try \ { \ XTESTS_INVOKE_TEST_CASE_FN_INNER_(fn); \ \ XTESTS_NS_C_QUAL(xtests_caseExceptionExpected)(#type); \ } \ catch(XTESTS_NS_CPP_QUAL(requirement_failed_exception)& /* x */) \ { \ ...

Build fails on GCC 4.6 in Ubuntu 11.10x64

Abe Bachrach — Sun, 20 Nov 2011 00:25:59 -0000

I'm trying to compile fast format 0.7.1-alpha-3 on ubuntu 11.10x64 which uses GCC4.6

gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/4.6.1/lto-wrapper
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Ubuntu/Linaro 4.6.1-9ubuntu3' --with-bugurl=file:///usr/share/doc/gcc-4.6/README.Bugs --enable-languages=c,c++,fortran,objc,obj-c++,go --prefix=/usr --program-suffix=-4.6 --enable-shared --enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --with-gxx-include-dir=/usr/include/c++/4.6 --libdir=/usr/lib --enable-nls --with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-plugin --enable-objc-gc --disable-werror --with-arch-32=i686 --with-tune=generic --enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu --target=x86_64-linux-gnu
Thread model: posix
gcc version 4.6.1 (Ubuntu/Linaro 4.6.1-9ubuntu3)

I built in the gcc44 target because there isn't one for 4.6

unfortunately the build fails on:

g++ -c -Wall -pedantic -Wno-long-long -Wno-unused-value -Wno-non-virtual-dtor -Werror -D_REENTRANT -D_DEBUG -UNDEBUG -Dunix -DFASTFORMAT_NO_IMPLICIT_LINK -I../../include -I/tmp/ff/stlsoft-1.9.109/include -o test.component.FILE.mt.debug.o ../../test/component/test.component.FILE/test.component.FILE.cpp
In file included from /tmp/ff/stlsoft-1.9.109/include/unixstl/filesystem/memory_mapped_file.hpp:66:0,
from /tmp/ff/stlsoft-1.9.109/include/platformstl/filesystem/memory_mapped_file.hpp:73,
from /tmp/ff/stlsoft-1.9.109/include/platformstl/filesystem/file_lines.hpp:72,
from ../../test/component/test.component.FILE/test.component.FILE.cpp:32:
/tmp/ff/stlsoft-1.9.109/include/unixstl/filesystem/filesystem_traits.hpp: In static member function ‘static stlsoft::unixstl_project::filesystem_traits<char>::bool_type stlsoft::unixstl_project::filesystem_traits<char>::create_directory(const char_type*)’:
/tmp/ff/stlsoft-1.9.109/include/unixstl/filesystem/filesystem_traits.hpp:1219:21: error: variable ‘mode’ set but not used [-Werror=unused-but-set-variable]
/tmp/ff/stlsoft-1.9.109/include/unixstl/filesystem/filesystem_traits.hpp: In static member function ‘static stlsoft::unixstl_project::filesystem_traits<char>::file_handle_type stlsoft::unixstl_project::filesystem_traits<char>::invalid_file_handle_value()’:
/tmp/ff/stlsoft-1.9.109/include/unixstl/filesystem/filesystem_traits.hpp:1282:16: error: converting to non-pointer type ‘stlsoft::unixstl_project::filesystem_traits<char>::file_handle_type {aka int}’ from NULL [-Werror=conversion-null]
cc1plus: all warnings being treated as errors

By disabling -Wall the compilation gets a bit futher, before running into:

g++ -c -Werror -D_REENTRANT -D_DEBUG -UNDEBUG -Dunix -DFASTFORMAT_NO_IMPLICIT_LINK -I../../include -I/tmp/ff/stlsoft-1.9.109/include -o test.component.FILE.mt.debug.o ../../test/component/test.component.FILE/test.component.FILE.cpp
In file included from /tmp/ff/stlsoft-1.9.109/include/unixstl/filesystem/memory_mapped_file.hpp:66:0,
from /tmp/ff/stlsoft-1.9.109/include/platformstl/filesystem/memory_mapped_file.hpp:73,
from /tmp/ff/stlsoft-1.9.109/include/platformstl/filesystem/file_lines.hpp:72,
from ../../test/component/test.component.FILE/test.component.FILE.cpp:32:
/tmp/ff/stlsoft-1.9.109/include/unixstl/filesystem/filesystem_traits.hpp: In static member function ‘static stlsoft::unixstl_project::filesystem_traits<char>::file_handle_type stlsoft::unixstl_project::filesystem_traits<char>::invalid_file_handle_value()’:
/tmp/ff/stlsoft-1.9.109/include/unixstl/filesystem/filesystem_traits.hpp:1282:16: error: converting to non-pointer type ‘stlsoft::unixstl_project::filesystem_traits<char>::file_handle_type {aka int}’ from NULL [-Werror=conversion-null]
cc1plus: all warnings being treated as errors

if I disable -Werror, the build completes.

For reference, things built just fine on ubuntu 10.10x64 (which uses gcc 4.4.5)

thanks!

Wrong behaviour of fmtln (linux).

Anonymous — Tue, 29 Mar 2011 15:48:21 -0000

In example.overload.3.floatingpoint_columns.cpp if one try to output with "[{0,14,3,<}{1,14,3,<}]" such format string (line 213) the result is [123123]. gcc 4.5.2 Linux

Strange behaviour on Mac Tiger

Zara — Thu, 25 Nov 2010 09:28:30 -0000

The preprocessing of the header files in Mac Tiger (10.4), removes last 'r' from lines, i.e.
struct xtests_integer_failure_reporter_selector
is converted into:
struct xtests_integer_failure_reporter_selecto
giving all kind of problems when compiling

Inconsistent conditionals in namespace suppression

Matt Wilson — Wed, 10 Nov 2010 21:59:24 -0000

In files:

* src/api.cpp
* src/fmt_cache.cpp
* src/fmt_spec_defect_handlers.cpp
* src/memory_pool.cpp
* src/replacements.cpp

See https://sourceforge.net/projects/fastformat/forums/forum/612782/topic/3939419 for further details

CComBSTR sink fails when adding 0-length to !empty instance

Matt Wilson — Tue, 14 Sep 2010 03:23:02 -0000

- need to add this test-case to all sink tests

Max width of 0 is not respected

Matt Wilson — Mon, 17 May 2010 05:58:05 -0000

Is ignored. Should truncate to empty string