[FaTony]

WinXP SP3, VS 2010.

PHP Code:

const char *HandleTypes[] =
{
   "ABC",
   "DEF"
   //etc
}
void Foo(void)
{
    //According to debugger:
    //HandleTypes[0] type is "char *"
    //HandleTypes[0] value is "ABC"
    handlesys->CreateType(HandleTypes[0], ...); //HandleError_Parameter
} 


PHP Code:

void Foo(void)
{
    const char *HandleTypes[] =
     {
        "ABC",
        "DEF"
        //etc
     }
    //According to debugger:
    //HandleTypes[0] type is "const char *"
    //HandleTypes[0] value is "ABC"
    handlesys->CreateType(HandleTypes[0], ...); //No errors
} 


What the hell? Why?

[pRED*]

Of the 3 possible reasons that error could occur, only attempting to register an already existing name seems relevant.

The given code example looks to be fine, sure you didn't inadvertently fix another issue while changing that?

[FaTony]

No, it seems to be how Windows manages memory. You see that debugger identify same variables as different types based on the scope. And handle type names I use in my extension are rather unique. What are the other 2 possible reasons?

[Fyren]

SM is open source, you can figure that out yourself by searching CreateType for HandleError_Parameter.

It will set err to HandleError_Parameter if the dispatch parameter you gave is null, you give a bad parent type, or you register the same name more than once.

[FaTony]

You are right. I guess I'll wait for the next stable SM build and compile it myself to see everything through debugger. I'm currently on snapshot and the source code I have does not match my binaries.

[BAILOPAN]

Full line of code would be good, another parameter could be at fault

[FaTony]

http://forums.alliedmods.net/showthread.php?t=146674

SPVTypeHandler.cpp

There are both cases in there. Just comment local array and uncomment global.

[PM]

Hm this is weird.
Does this happen with all of your 3 handle types or only one (if yes, which one) of them?

Maybe try something like

printf("%d: %x %s\n", i, SPVTypeNames[i], SPVTypeNames[i]);
typebuffer[i].type = handlesys->CreateType(...);
printf("%d: %x %s\n", i, SPVTypeNames[i], SPVTypeNames[i]);
printf("Result: %d\n\n", typebuffer[i].type);

(replace printf with some other printing/logging mechanism that you can see)

That way you can see:
- what exactly was passed into CreateType (address of name and string value stored on that address)
- if something modifies the values in the process

As far as the code SM goes, I don't see any reason why it would do this

[FaTony]

Ok, will try in a couple of days.

[FaTony]

Quite strange, that issue seems to have been resolved by itself. Must've been a snapshot.