[
Lists Home |
Date Index |
Thread Index
]
> Section 4.3.11 (as corrected) reads:
Thank you for the clarifications. Is there a way to get the corrected specification,
not a separate file with corrections?
Is it possible to check with the errata each time one feels there is a contradiction;
but this is not very time efficient, because I feel there is a contradiction too often.
- Is NaN equal to itself? Is NaN less than +INF and greater than -INF? If
+INF and -INF are exclusive restrictions, do they fall within the range of their own?
- For decimal, leading and trailing zero digits are optional. This is not the
same as 'leading and fractional trailing zero digits', although I understand
that the authors meant the latter. If they would say what they meant, they could
also add "including decimal point".
But still, the definition is probably wrong, since not all leading zero digits are
optional, one should be kept unless there is something in the fractional part.
Or is '-.' a valid decimal?
Using a simple regular expression to specify the lexical space would be a much easire way
to understand it.
And since the specification cannot be used to implement the datatypes:
Is there a reference implementation that parses built-in types right?
Is there a test suite that actually shows what the authors meant?
David Tolpin
|
