Fix a narrowing conversion #964
Conversation
| @@ -205,19 +205,19 @@ extern int RESFACTOR; | |||
| ** Enable the set of limited cheat key options. | |||
| */ | |||
| #ifdef VIRGIN_CHEAT_KEYS | |||
| #define PARM_PLAYTEST 0xF7DDC227 // "PLAYTEST" | |||
| #define PARM_PLAYTEST 0xF7DDC227UL // "PLAYTEST" | |||
There was a problem hiding this comment.
uint32_t is not unsigned long on all platforms.
| @@ -165,7 +165,7 @@ class xTargetClass | |||
| void Invalidate(void) | |||
| { | |||
| Target.Sub.Exponent = RTTI_NONE; | |||
| Target.Sub.Mantissa = -1; | |||
| Target.Sub.Mantissa = (1UL << TARGET_MANTISSA) - 1; | |||
There was a problem hiding this comment.
I don't think this is equivalent?
| @@ -2318,7 +2318,7 @@ void Load_Recording_Values(void) | |||
| * HISTORY: * | |||
| * 08/19/1995 JLB : Created. * | |||
| *=============================================================================================*/ | |||
| long Obfuscate(char const* string) | |||
| uint32_t Obfuscate(char const* string) | |||
There was a problem hiding this comment.
The long type needs to go as its width varies between platforms but it should just change to int to match the ra code?
|
It's not only to get rid of the warning. Obfuscate() currently operates on an int, but in TD returns a long (unlike RA, which currently returns an int). Comparing that to a constant like 0xF7DDC227 may fail if the function actually returned 0xFFFFFFFFF7DDC227 because it was sign-extended to long. The only safe way is to declare them to return unsigned int (or uint32_t in this case), since this is what they actually calculate. |
|
I'd be happy to see it changed to unsigned int in both cases. |
No description provided.