Uninitialized members and Typeinfo

[nbohr1more]

@duzenko

Linux build now fails:

 

[ 73%] Building CXX object CMakeFiles/TheDarkMod.dir/game/gamesys/TypeInfo.cpp.o
/home/randy/tdm_src/trunk/game/gamesys/TypeInfo.cpp:51:10: fatal error: GameTypeInfo.h: No such file or directory
 #include "GameTypeInfo.h"    // Make sure this is up to date!
          ^~~~~~~~~~~~~~~~
compilation terminated.
CMakeFiles/TheDarkMod.dir/build.make:5252: recipe for target 'CMakeFiles/TheDarkMod.dir/game/gamesys/TypeInfo.cpp.o' failed
make[2]: *** [CMakeFiles/TheDarkMod.dir/game/gamesys/TypeInfo.cpp.o] Error 1
make[2]: *** Waiting for unfinished jobs....
In file included from /home/randy/tdm_src/trunk/idlib/../idlib/Lib.h:236:0,
                 from /home/randy/tdm_src/trunk/build/TheDarkMod_pch/precompiled.h:45:
/home/randy/tdm_src/trunk/idlib/../idlib/Str.h:815:36: warning: argument 1 null where non-null expected [-Wnonnull]
     int l = static_cast<int>(strlen(text));
                              ~~~~~~^~~~~~
In file included from /home/randy/tdm_src/trunk/ThirdParty/artefacts/tracy/include/Tracy.hpp:102:0,
                 from /home/randy/tdm_src/trunk/build/TheDarkMod_pch/precompiled.h:19:
/usr/include/string.h:384:15: note: in a call to function ‘size_t strlen(const char*)’ declared here
 extern size_t strlen (const char *__s)
               ^~~~~~
At global scope:
cc1plus: warning: unrecognized command line option ‘-Wno-class-memaccess’
CMakeFiles/Makefile2:94: recipe for target 'CMakeFiles/TheDarkMod.dir/all' failed
make[1]: *** [CMakeFiles/TheDarkMod.dir/all] Error 2
Makefile:102: recipe for target 'all' failed
make: *** [all] Error 2

 

[duzenko]

Sorry, reverted that

[stgatilov]

How did you get away without #define private public ?

As far as I remember, it is required to get member pointers.

[duzenko]

9 minutes ago, stgatilov said:

How did you get away without #define private public ?

As far as I remember, it is required to get member pointers.

I have no idea!11

I can see that text save code calls class methods to get members, could it be it? No idea how it works with sizeof/offsetof  though, maybe those are immune to access modifiers?

[duzenko]

On 12/16/2021 at 2:58 PM, stgatilov said:

Yes, padding is a problem.

One possible thing is to check every 8 bytes instead of 4 bytes: I guess it is already done like that in 64-bit builds. Of course, it allows the check to miss uninitialized members, but I don't think there is a way around it.

In alternative reality I can parse pdb's and get all type info from there

[stgatilov]

5 minutes ago, duzenko said:

In alternative reality I can parse pdb's and get all type info from there

I think one day I'll get member offsets using libclang.
And I'm pretty sure typeinfo will be removed soon... not even sure which will happen earlier.