Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Radek]

When trying a "make check" with today's snapshot I get the following

Code: [Select]

if g++ -DHAVE_CONFIG_H -I. -I. -I../..    -I. -I./.. -I../../src -I../../src/libs -I/mingw/lib/wx/include/msw-unicode-release-static-2.7 -I/mingw/include/wx-2.7 -D__WXMSW__ -mthreads -DwxUSE_GUI=0 -DMULEUNIT    -DNOMINMAX -DUSE_EMBEDDED_CRYPTO -W -Wall -Wshadow -Wundef -g -ggdb -fno-inline -D__DEBUG__ -fmessage-length=0 -D_GLIBCXX_DEBUG -D_GLIBCXX_DEBUG_PEDANTIC   -MT FileDataIOTest-FileDataIOTest.o -MD -MP -MF ".deps/FileDataIOTest-FileDataIOTest.Tpo" -c -o FileDataIOTest-FileDataIOTest.o `test -f 'FileDataIOTest.cpp' || echo './'`FileDataIOTest.cpp; \
then mv -f ".deps/FileDataIOTest-FileDataIOTest.Tpo" ".deps/FileDataIOTest-FileDataIOTest.Po"; else rm -f ".deps/FileDataIOTest-FileDataIOTest.Tpo"; exit 1; fi
In file included from ../../src/CFile.h:29,
                 from FileDataIOTest.cpp:2:
../../src/SafeFile.h:204:2: warning: #warning DOCS
FileDataIOTest.cpp: In member function `virtual void CMemFileAttachedBufferTest::run()':
FileDataIOTest.cpp:648: error: call of overloaded `CMemFile(NULL, int)' is ambiguous
../../src/MemFile.h:93: note: candidates are: CMemFile::CMemFile(const uint8_t*, size_t)
../../src/MemFile.h:92: note:                 CMemFile::CMemFile(uint8_t*, size_t)
make[3]: *** [FileDataIOTest-FileDataIOTest.o] Error 1

[Kry]

Hah. My changes PWNED Xaignar.

[Xaignar]

It's Kry's fault!

[Kry]

No it isn't!

[Radek]

With today's snapshot I get the following

Code: [Select]

Running test-collection "RangeMap" with 42 test-cases:
        Test "DefaultConstructor"
                Failure: "Expected '[(100, 150, 0)]' but got '[(100, 0, 150)]'" line 85 in RangeMapTest.cpp
        Test "CopyConstructor"
FAIL: RangeMapTestThis was after I had received a popup from RangeMapTest.exe with the message, that an instruction at 0x004b263e would try to read from memory location 0x00000008, which was not possible.

Code: [Select]

================================
1 of 5 tests failed
Please report to [EMAIL]admin@amule.org[/EMAIL]
================================

[Kry]

haha!

[Xaignar]

Please test tomorrows snapshot, or apply this patch and tell me what happens. It seems to me that something rather strange is going on there.

[Radek]

No difference with today's snapshot (20060126) ?!

For the records: gcc 3.4.5, MingW, wxCVS, WinXP.
And this time I made an extra start of RangeMapTest.exe with gdb

Code: [Select]

#0  0x004b25d6 in std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_root() const (this=0x0)
    at e:/msys/mingw/bin/../lib/gcc/mingw32/3.4.5/../../../../include/c++/3.4.5/bits/stl_tree.h:442
#1  0x004b3968 in std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_Rb_tree(std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > > const&) (this=0x22fc10, __x=@0x0)
    at e:/msys/mingw/bin/../lib/gcc/mingw32/3.4.5/../../../../include/c++/3.4.5/bits/stl_tree.h:557
#2  0x004982a0 in __gnu_norm::map, std::less, std::allocator > > >::map(__gnu_norm::map, std::less, std::allocator > > > const&) (this=0x22fc10, __x=@0x0)
    at e:/msys/mingw/bin/../lib/gcc/mingw32/3.4.5/../../../../include/c++/3.4.5/bits/stl_map.h:166
#3  0x0049f4e8 in __gnu_debug_def::map, std::less, std::allocator > > >::map(__gnu_debug_def::map, std::less, std::allocator > > > const&) (this=0x22fc10,
    __x=@0x0)
    at e:/msys/mingw/bin/../lib/gcc/mingw32/3.4.5/../../../../include/c++/3.4.5/debug/map.h:86
#4  0x004acee0 in CRangeMap::CRangeMap(CRangeMap const&) (this=0x22fc10, other=@0x0)
    at ../../src/RangeMap.h:239
#5  0x004017fc in RangeMapCopyConstructorTest::run() (this=0x4cf070)
    at RangeMapTest.cpp:186
#6  0x0048bd8f in muleunit::TestCase::run() (this=0x3d63d0) at testcase.cpp:111
#7  0x0048b69a in muleunit::TestRegistry::runTests() (this=0x4d66c0)
    at testregistry.cpp:81
#8  0x0048b329 in muleunit::TestRegistry::runAndPrint() ()
    at testregistry.cpp:52
#9  0x00498ee3 in UnitTestApp::OnRun() (this=0x3d8fa0) at main.cpp:36
#10 0x0047cdd7 in wxEntryReal(int&, wchar_t**) ()
#11 0x00489ad0 in main (argc=1, argv=0x3d2d98) at main.cpp:283
(gdb) cont
Continuing.

Program exited with code 030000000005.


[Xaignar]

Radek,
could you try inserting the following snipped in unittests/tests/RangeMapTest.cpp:

Old:

Code: [Select]

TEST(RangeMap, DefaultConstructor)
{
TestRangeMap map;
[snip]
New:

Code: [Select]

TEST(RangeMap, DefaultConstructor)
{
uint64 foo = 100, bar = 150;
int* baz = (int*)1;
wxString teststr = wxString::Format(wxT("%llu %llu %li), foo, bar, baz);

TestRangeMap map;
[snip]
And tell me the output of running the test.


EDIT: Also, if possible, can you join #amule? Since that would speed up this exchange a bit.

[ken]

Uh.  OK, I made the suggestion to Xaignar that he make this suggestion to you, but apparently I wasn't clear.  Or Xaignar was sleepy.

The new code should be:

Code: [Select]

TEST(RangeMap, DefaultConstructor)
{
uint64 foo = 100, bar = 150;
int* baz = (int*)1;
wxString teststr = wxString::Format(wxT("%llu %llu %li"), foo, bar, baz);
ASSERT_EQUALS(wxT("100 150 1"), teststr);

TestRangeMap map;
[snip]
The difference being: 1) corrected a missing quote character, and 2) actually test the value of teststr.  

[Radek]

I ran "make check" with the patch above. This is the output

Code: [Select]

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

Running test-collection "RangeMap" with 42 test-cases:
        Test "DefaultConstructor"
                Failure: "Expected '[(100, 150, 0)]' but got '[(100, 0, 150)]'" line 91 in RangeMapTest.cpp
                Failure: "Expected '100 150 1' but got '100 0 150'" line 178 in RangeMapTest.cpp
        Test "CopyConstructor"
                Failure: "Expected '[(100, 150, 0)]' but got '[(100, 0, 150)]'" line 91 in RangeMapTest.cpp
        Test "AssignmentOperator"
                Failure: "Expected '[(100, 150, 0)]' but got '[(100, 0, 150)]'" line 91 in RangeMapTest.cpp
        Test "Equality"
                Failure: "Expected '[(100, 150, 0)]' but got '[(100, 0, 150)]'" line 91 in RangeMapTest.cpp
                Failure: "mapA == mapB" line 238 in RangeMapTest.cpp
        Test "Iterators"
                Failure: "Expected '[(100, 150, 0)]' but got '[(100, 0, 150)]'" line 91 in RangeMapTest.cpp
                Failure: "Expected '(100, 124, 0)' but got '(125, 0, 175)'" line 260 in RangeMapTest.cpp
        Test "Erase"
                Failure: "Expected '[(100, 150, 0)]' but got '[(100, 0, 150)]'" line 91 in RangeMapTest.cpp
                Failure: "Expected '[(100, 124, 0), (125, 175, 2), (176, 200, 1)]' but got '[(125, 0, 175)]'" line 290 in RangeMapTest.cpp
        Test "Clear"
                Failure: "Expected '[(100, 150, 0)]' but got '[(100, 0, 150)]'" line 91 in RangeMapTest.cpp
        Test "FindRange"
e:/msys/mingw/bin/../lib/gcc/mingw32/3.4.5/../../../../include/c++/3.4.5/debug/safe_iterator.h:193:
    error: attempt to dereference a past-the-end iterator.

Objects involved in the operation:
iterator "this" @ 0x0022FBE0 {
type = N11__gnu_debug14_Safe_iteratorISt17_Rb_tree_iteratorISt4pairIKy15CRangeMapHelperIPiyEEEN15__gnu_debug_def3mapIyS6_St4lessIyESaIS7_EEEEE (mutable iterator);
  state = past-the-end;
  references sequence with type `N15__gnu_debug_def3mapIy15CRangeMapHelperIPiyESt4lessIyESaISt4pairIKyS3_EEEE' @ 0x0022FBE0
}
FAIL: RangeMapTestNo popup anymore.

[ken]

Failure: "Expected '100 150 1' but got '100 0 150'" line 178 in RangeMapTest.cpp

This confirms that the problem is not due to any aMule code.  It's an interaction between wxString::Format, the compiler, and the C runtime library.  It apparently doesn't cope properly with 64-bit arguments to vararg functions.  Or, the particular variant of printf which is used inside of wxString::Format doesn't cope well with "%llu" format specifiers.  Or, it's a weird hack that wxString::Format does: it copies a va_list data structure, which seems to be an unsupported operation and might not be working properly.

[Radek]

@ken
In the meantime we made some experiments and found out, that it was indeed a problem with the format specifier "%llu". It was not interpreted as 64bit, so the argument was taken as 32bit, the next argument from lower half of the first argument and so on

When compiling under MingW/Win32, the format specifiers of some M$-DLL have to be used, because the call of *printf finally goes there. For these DLLs the correct specifier would be "%I64u", which I personally consider very ugly (the typical "special" way of M$).

There is a workaround (a DEFINE in defs.h) for this, which is used now in the test suite.

Starting with yesterday's snapshot, the problem is solved. I just forgot to change the thread title...