Switched to reserved identifiers.
================
Comment at: include/experimental/algorithm:53
@@ +52,3 @@
+ for (; __first != __last; ++__first, (void)++__k) {
+ _Distance __r = _VSTD::uniform_int_distribution<>(0, __k)(__g);
+ if (__r < __sz)
----------------
This thing is costing us lots of performance - it's just way too fancy.
It makes this code is about 2x slower than std::random_sample in existing
implementations, but I guess it makes the sample more uniform, as well.
Do you thing smth like (__g() - __g::min()) % ( __k + 1) would be acceptable?
================
Comment at: include/experimental/algorithm:49
@@ +48,3 @@
+ Distance k = 0;
+ for (; first != last && k < n; ++first, ++k)
+ out[k] = *first;
----------------
EricWF wrote:
> We need a void cast like `++first, (void) ++k` to prevent calling a comma
> operator overload.
done
http://reviews.llvm.org/D9044
EMAIL PREFERENCES
http://reviews.llvm.org/settings/panel/emailpreferences/
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
