Dan Hoeflinger

> Should we think about ability to save previous behavior when the macros `TEST_LONG_RUN` is defined? I've now made it so when `TEST_LONG_RUN` is true, we undo any *reduction* in...

Actually... it looks like TEST_LONG_RUN=1 is used in the nightly testing, so it kind of defeats the purpose of this PR to turn off the savings in this case if...

> Actually... it looks like TEST_LONG_RUN=1 is used in the nightly testing, so it kind of defeats the purpose of this PR to turn off the savings in this case...

> I'm tempted to undo my TEST_LONG_RUN changes so we can take advantage of the savings in the nightly suite. I've reverted this change, because the point is to provide...

I agree that the values here shouldn't need to be atomics, and the atomics for the status flags can provide everything we need.

I'm working on adding some unit testing for tuple to the test suite. Once I add that, I will probably revert the changes to `zip_iterator.pass` .

> Based on this, it seems like the three functions that are defined are sufficient. Yes, thanks for doing this enumeration of the possibilities. Working on a test which should...

OK, now all those options @adamfidel mentioned should be covered by the unit test. I've reverted the `zip_iterator` test changes in favor of unit testing. I've also rebased to have...

> I think better if we will support all restrictions for these operators like in https://en.cppreference.com/w/cpp/utility/tuple : all these operators will be removed in C++20. @SergeyKopienko Can you be more...

> For me it's looks like we should implement all these operators only for C++17 version and implement `` operator for C++20. That's fine. I can do that.