What is this BS?

[ZerGreenOne]

if (0 - 0 == 0) then
    call BJDebugMsg("1")
else
    call BJDebugMsg("2")
endif

if (0 - 0 != 0) then
    call BJDebugMsg("1")
else
    call BJDebugMsg("2")
endif

if (0 - 0.0 == 0) then
    call BJDebugMsg("a")
else
    call BJDebugMsg("b")
endif

if (0 - 0.0 != 0) then
    call BJDebugMsg("a")
else
    call BJDebugMsg("b")
endif

Using this code I get the following output:

1
2
a
a

Could someone tell me why? Is this a bug in the world editor or is it working as intended? Also, I find it annoying how dividing by 0 stops execution of the script rather than just being ignored or outputting 0 or NaN, since there is no way to handle errors.

It's super annoying, how can I determine whether (some real variable that could be equal to 0 subtracted from another real variable that could be equal to 0) is equal to 0 without it thinking that 0.1 is equal to 0?

It only occurs when 0 is subtracted from 0, and the subtractor is a real. if either is non-zero than it will work correctly. But I don't want to have to check whether one is non-zero, in fact the whole purpose of this is to get around the fact that dividing by 0 will stop execution, and I do this by checking if the divisor is equal to 0. I don't want yet another step to check whether one part of the divisor is equal to 0. Maybe I could instead check whether one is equal to the other, but it's still super annoying and buggy.

I think I can get around this now, by using:

if (1 != 0.0) then
call BJDebugMsg("a")
else
call BJDebugMsg("b")
endif

if (1 == 0.0) then
call BJDebugMsg("a")
else
call BJDebugMsg("b")
endif

But it is still super annoying and buggy, and it made me spend a while trying to fix my perfectly OK code.

Actually, I fixed my problem, maybe someone would like to move this thread to patch discussion or something (as a bug report) or even delete it.

 

[DracoL1ch]

!= compares things byte-by-byte.
statement 0. != 0 can be true as well
== compares things much less sharpy
since you mess with typecasting real & integer, you receive this resuit. use not a==b in case if you wanna operate floats

 

[DracoL1ch]

there was article about it somewhere at hive.

damn, its a new engine but doubleposting doesnt stick up yet. what year is it?

 

[Wietlol]

2081?

Also, I thought that (according to the thread you probably mentioned) != was actually better to use for accuracy...
But yea, the meaning of certain bits inside integers mean something else than the ones in floating points, which is probably the problem.

0. != 0
0. == 0.