| Add Proposal | Add Analysis | Edit Class, Environment, or Release |
| Number | 664
|
| Notify-List |
|
| Category | errata
|
| Synopsis | Missing rule for reduction operators in 4.5.1
|
| State | open
|
| Class | errata-simple
|
| Arrival-Date | Apr 22 2005
|
| Originator | sharp@cadence.com (Behavioral Task Force)
|
| Release |
|
| Environment |
|
| Description |
The rules for expression types in 4.5.1 do not address the type of the result of a reduction operator (such as reduction-OR). Since the operand is self-determined, the general rule about non-self-determined operands does not apply. All other operators that have only self-determined operands produce an unsigned result. The other operators that produce a 1-bit truth value independent of operand size (i.e. the compare operators) produce an unsigned result. The original implementation in XL from which the signed arithmetic proposal was derived produces an unsigned result. It seems clear that the result of a reduction operator should be unsigned. |
| Fix |
In 4.5.1, CHANGE "Comparison results (1, 0) are unsigned, regardless of the operands." TO "Comparison and reduction operator results (1, 0) are unsigned, regardless of the operands." |
| Unformatted |
|
Hosted by Boyd Technology