Version 2.11, Proposal for interval standardization
Earlier (at 15:23 UTC) I wrote:
| There is a reproducibility issue here however -- but I think that P1788
| would be ill-advised to recommend, never mind mandate (under special mode
| perhaps), reproducibility across implementations.
I want to take some of that back. It *is* useful to be able to specify
explicitly an order of evaluation, and most languages do provide such
means, e.g. by use of parentheses or perhaps double parentheses, so as
to distinguish this mode from the one where the compiler is free to
rearrange the sequence, perform strength reductions etc.
In particular, the example (1/interval(3))*3 could perhaps be used
deliberately to explore the resolution of the IA implementation type.
Michel.
Sent: 2008-11-18 17:09:49 UTC