| ??? 01/13/11 08:35 Read: times |
#180544 - works Responding to: ???'s previous message |
The new version compiles ((char *)0)[0x10] = 0; without a single complaint.
This is a bit strange, as ((char *)1)[0x10] = 0; warns about "cast of LITERAL value to 'generic' pointer". I'd say this would be appropriate for the null constant cast to non-(void *) (i.e. creating null pointer) too. I also maintain my objections to using null pointer as a base for "indexing" (array subscripting). C99 in 6.5.2.1 requires "one of the extensions shall have type "pointer to object type"*; while the status of null pointer is IMHO that it is not a pointer to object. I know this is hair-splitting and PITA, but once we are discussing it... :-) Jan --- * Btw., SDCC violates the standard by not supporting the "reverse" array subscripting, e.g. 0x10[((char *)1)], however perverse that idea is; but I am not going to put this as a feature request *exactly* because it IS perverse. |
| Topic | Author | Date |
| sdcc internal error / C syntax | 01/01/70 00:00 | |
| legality of indexing NULL pointer | 01/01/70 00:00 | |
| bug | 01/01/70 00:00 | |
| version | 01/01/70 00:00 | |
| version revisited | 01/01/70 00:00 | |
| thank you | 01/01/70 00:00 | |
| Fixed | 01/01/70 00:00 | |
| no snapshot | 01/01/70 00:00 | |
| works | 01/01/70 00:00 | |
| 0 is special - but so is NULL. indexing around NULL is bad | 01/01/70 00:00 | |
| No guarantee that a NULL pointer points to any memory | 01/01/70 00:00 | |
| time | 01/01/70 00:00 | |
| I know :-( | 01/01/70 00:00 | |
| Use of __at ? | 01/01/70 00:00 | |
| What is "that way"? | 01/01/70 00:00 | |
| the antique version.... | 01/01/70 00:00 | |
| XBYTE macro | 01/01/70 00:00 | |
| Okay, then the following definition... | 01/01/70 00:00 | |
| RE: David's remarks about volatility | 01/01/70 00:00 | |
| Close, but no cigar | 01/01/70 00:00 | |
| Avoid the 'volatile' | 01/01/70 00:00 | |
| instead of offsetting... | 01/01/70 00:00 | |
| Dereferencing a '_REG | 01/01/70 00:00 | |
I agree | 01/01/70 00:00 | |
| Use of __at ? [ed] | 01/01/70 00:00 |