The latest SVN bugs thread

[CuteAlien]

greenya: Yes, things added to trunk will in most cases not be backported to 1.7, think of trunk as 1.8. The 1.7 branch is just receiving bugfixes.
And yeah - we haven't released a major version in a long time, we're working on getting it out asap now.

[Nadro]

08.SpecialFx + D3D9 + 'y' mode -> shadow isn't visible (in a rev3946 is ok) and also particles are incorrectly rendered (black areas, also visible in a rev3750).

[CuteAlien]

I've run tests today on different platforms, this are the current results (all debug, more tests fail in release):

Failing test results (24.3.2011) VS 2010 Windows 7 OS 64-bit compiling for 32-bit: 47, 49, 50, 51, 52, 53, 55, 57
Failing test results (26.3.2011) GCC Ubuntu compile 64-bit: 19, 22, 44, 47, 48, 49, 50, 51, 52, 54, 55, 56, 57, 58, 59, 60, 63, 64
Failing test results (26.3.2011) GCC Debian 32-bit: 47, 48, 49, 50, 51, 52, 54, 55, 56, 57, 58, 59, 60, 63, 64
Failing test results (26.3.2011) VS 2010 Windows XP 32-bit: 47, 49, 50, 51, 52, 53, 54, 55, 57,

So rather obvious we have to do better. But also maybe as info why other things currently have a hard time getting in... we shouldn't really add new stuff while our test-suite is broken as we don't see anymore then if new changes break even more things (and yeah, I also still add new stuff sometimes *sigh*, but I try to be careful not making the situation worse).

Anyone who can help there - for example figuring out which tests got broken in which svn revision would often help - is welcome (especially on 32-bit platforms, some tests probably never worked on 64-bit so far, I'm working mostly on those currently). If you want to help figuring out breaking revisions, you can do so easily - just run tests on your system, note which work (in tests.log) and then start checking out older revisions and compare until you have the revision where something changes. The test-suite is in folder "tests" (yeah.... hard to find *g*). And certainly if you can even find out _why_ a test breaks, what it's supposed to do and how it should be fixed... oh well I guess I'm just dreaming there ... but that sort of info would be great :-) (Note that in many cases tests probably just didn't get updated on engine-changes)

edit (2012/04/03): increased all testnumbers by one on "GCC Ubuntu compile 64-bit" as I noticed test 8 was outcommented to get it compiling. Should probably better replace numbers by names anyway (not now).

[wing64]

Wrong variable
File: CSceneNodeAnimatorFlyCircle.cpp
Function: CSceneNodeAnimatorFlyCircle::serializeAttributes()
Line: 66

old

Code: Select all

out->addFloat("RadiusEllipsoid", Radius);

new

Code: Select all

out->addFloat("RadiusEllipsoid", RadiusEllipsoid);

[CuteAlien]

@wing64: Thanks. That was already wrong in 1.7 so I fixed it there, will be fixed in trunk with next merge.

[wing64]

In example 15.LoadIrrFile why sphere have not root is cube ?
I saw in xml <node type="sphere"> is child of <node type="cube"> and i try load in irredit1.5 for checked result was right.

[Randajad]

Irrlicht failed to compiling with MinGW 4.6.4.

COpenGLDriver.cpp:159:2: error: narrowing conversion of '(((irr::video::COpenGLDriver*)this)->irr::video::COpenGLDriver::Params.irr::SIrrlichtCreationParameters::Stencilbuffer ? 1 : 0)' from 'int' to 'BYTE {aka unsigned char}' inside { } [-fpermissive]

Code: Select all

        PIXELFORMATDESCRIPTOR pfd = {
                sizeof(PIXELFORMATDESCRIPTOR),             // Size Of This Pixel Format Descriptor
                1,                                         // Version Number
                PFD_DRAW_TO_WINDOW |                       // Format Must Support Window
                PFD_SUPPORT_OPENGL |                       // Format Must Support OpenGL
                (Params.Doublebuffer?PFD_DOUBLEBUFFER:0) | // Must Support Double Buffering
                (Params.Stereobuffer?PFD_STEREO:0),        // Must Support Stereo Buffer
                PFD_TYPE_RGBA,                             // Request An RGBA Format
                Params.Bits,                               // Select Our Color Depth
                0, 0, 0, 0, 0, 0,                          // Color Bits Ignored
                0,                                         // No Alpha Buffer
                0,                                         // Shift Bit Ignored
                0,                                         // No Accumulation Buffer
                0, 0, 0, 0,                                    // Accumulation Bits Ignored
                Params.ZBufferBits,                        // Z-Buffer (Depth Buffer)
                Params.Stencilbuffer ? 1 : 0,              // Stencil Buffer Depth /// IT'S ERROR HERE!
                0,                                         // No Auxiliary Buffer
                PFD_MAIN_PLANE,                            // Main Drawing Layer
                0,                                         // Reserved
                0, 0, 0                                    // Layer Masks Ignored
        };

Should be:

Code: Select all

        PIXELFORMATDESCRIPTOR pfd = {
                sizeof(PIXELFORMATDESCRIPTOR),             // Size Of This Pixel Format Descriptor
                1,                                         // Version Number
                PFD_DRAW_TO_WINDOW |                       // Format Must Support Window
                PFD_SUPPORT_OPENGL |                       // Format Must Support OpenGL
                (Params.Doublebuffer?PFD_DOUBLEBUFFER:0) | // Must Support Double Buffering
                (Params.Stereobuffer?PFD_STEREO:0),        // Must Support Stereo Buffer
                PFD_TYPE_RGBA,                             // Request An RGBA Format
                Params.Bits,                               // Select Our Color Depth
                0, 0, 0, 0, 0, 0,                          // Color Bits Ignored
                0,                                         // No Alpha Buffer
                0,                                         // Shift Bit Ignored
                0,                                         // No Accumulation Buffer
                0, 0, 0, 0,                                    // Accumulation Bits Ignored
                Params.ZBufferBits,                        // Z-Buffer (Depth Buffer)
                BYTE(Params.Stencilbuffer ? 1 : 0),              // Stencil Buffer Depth
                0,                                         // No Auxiliary Buffer
                PFD_MAIN_PLANE,                            // Main Drawing Layer
                0,                                         // Reserved
                0, 0, 0                                    // Layer Masks Ignored
        };

Also irrlicht work's in wrong way with gcc >= 4.7, because __attribute__((packed)) there works some different and offsets are obtained by false.

Code: Select all

        // byte-align structures
#if defined(_MSC_VER) ||  defined(__BORLANDC__) || defined (__BCPLUSPLUS__)
#       pragma pack( push, packing )
#       pragma pack( 1 )
#       define PACK_STRUCT
#elif defined( __GNUC__ )
#       define PACK_STRUCT      __attribute__((packed))
#else
#       error compiler not supported
#endif

Why elif for GCC needed? Gcc supports pragmas for a long time. So it should be simple:

Code: Select all

        // byte-align structures
#if defined(_MSC_VER) ||  defined(__BORLANDC__) || defined (__BCPLUSPLUS__) || defined( __GNUC__ )
#       pragma pack( push, packing )
#       pragma pack( 1 )
#else
#       error compiler not supported
#endif
 
...
 
#if defined(_MSC_VER) ||  defined(__BORLANDC__) || defined (__BCPLUSPLUS__) || defined( __GNUC__ )
#       pragma pack( pop, packing )
#endif
 

Sorry for my eng, i'm ru.

[hybrid]

In which way does the pack attribute work differently now?

[Randajad]

It's makes variables to 2 byte aligned, but doesn't remove space at all. Maybe it's mingw bug, but anyway gcc supports pragmas in long time, so i think there is no reason to keep "elif" for __GNUC__.

[Randajad]

Another one bug. When building 64 bit version of irrlicht:

source\Irrlicht\CColladaMeshWriter.cpp:793:35: error: cast from 'const irr::scene::IMesh*' to 'long int' loses precision [-fpermissive]
name += irr::core::stringw((long)mesh);

... and many other warnings about casting in this file.

[hybrid]

The casts warning do always happen, a bug ticket was already submitted.

[CuteAlien]

Yes, sorry, I hope I get to fixing the collada-stuff tomorrow (then again I have a long day ahead including having to go to the tax-office... so might take yet another week until I get to it).

[Randajad]

Not warning, but error.
Talking about warnings there a lot warnings, so i can post it.

[CuteAlien]

Hm, strange, why didn't I get an error for that recently?

Anyway, I'm working on setting up 64-bit build on Linux again (new system which still has some troubles... ).

[hendu]

@CuteAlien

That's on a Mac, appears longs are 32-bit there. On 64-bit linux they are 64 and so no loss of precision.