On Friday, 1 March 2013 at 06:33:42 UTC, deadalnix wrote:
struct GenericSentinelRange(R, Sentinel) {
R r;
@property auto front() {
return r.front;
}
void popFront() {
r.popFront();
}
@property empty() {
return r.front == Sentinel;
}
}
That's exactly what I wrote as well, when I played with the idea
before posting. Except I just called the sentinel constant S and
added 'enum sentinel = S;' -- not significant extra work.
We don't need a new type of range at all. You confuse
legitimate uses case for using a sentinel to terminate a range
and uses case where an actual sentinel range is needed.
I'm not confused. The use case for a sentinel range /type/ is
always going to be the same, and singular: automating legitimate
use cases for sentinels. So, in effect, an argument for one is
as good as for the other, when I'm only responding to the calls
for use case examples. My mistake was in not making it clear that
I was responding just to those, sorry.
You can live without, and guess what : if you use LDC (and
probably GDC) you'll get the performance boost Walter is
talking about.
Which presupposes that I *can* use LDC (or GDC). For many people,
it may not be an option. And last time I tried (which has been
months now, admittedly) I couldn't get LDC to work for me at all,
which is a shame since I'm actually very much interested in it.
Either way, reliance on implementation details is not a good
thing.
