Re: About coverity 440107 (keyimpl.cxx in registry)

On 03/22/2013 10:12 PM, julien2412 wrote:
> Indeed we have this switch:
>      622     switch (valueType)
>      623     {
>      624         case RG_VALUETYPE_NOT_DEFINED:
>      625             memcpy(value, pBuffer, valueSize);
>      626             break;
>      627         case RG_VALUETYPE_LONG:
>      628             readINT32(pBuffer, *((sal_Int32*)value));
>      629             break;
>      630         case RG_VALUETYPE_STRING:
>      631             readUtf8(pBuffer, (sal_Char*)value, valueSize);
>      632             break;
>      633         case RG_VALUETYPE_UNICODE:
>      634             readString(pBuffer, (sal_Unicode*)value, valueSize);
>      635             break;
>      636         case RG_VALUETYPE_BINARY:
>      637             memcpy(value, pBuffer, valueSize);
>      638             break;
>      639         case RG_VALUETYPE_LONGLIST:
>      640         case RG_VALUETYPE_STRINGLIST:
>      641         case RG_VALUETYPE_UNICODELIST:
>      642             memcpy(value, pBuffer, valueSize);
>      643             break;
>      644     }
>
> but some lines above:
>      604     if (valueType > 4)
>      605     {
>      606         return REG_INVALID_VALUE;
>      607     }
>
> I took a look, it's like this since initial import in 2000.
> Should if (valueType > 4) be removed or should we clean the >4 cases ?

The >4 cases are the RG_VALUETYPE_*LIST cases for which there are 
dedicated ORegKey::get*ListValue functions, so they can be removed from 
the switch.  Shudder.  (All this would go to the bin shortly anyway if 
it weren't for our beloved backwards compatibility...)

Stephan