Max Stoiber

All backtrace tests are fixed! Unsure why those were failing since the backtraces were correct, just slightly different in formatting. Anyways, all the tests pass now! :tada: @neumino I'm going...

Huh, some backtrace tests still fail on wercker even though they pass locally? 🤔

Even with this branch we're getting changefeeds dropping out: ``` ReqlServerError: The connection was closed before the query could be completed for: r.table("usersNotifications").changes({ includeInitial: false }).filter(r.row("old_val").eq(null).and(r.row("new_val").eq(null).not()).or(r.row("old_val")("entityAddedAt").ne(r.row("new_val")("entityAddedAt"))))("new_val") .filter({ userId: "gVk5mYwccUOEKiN5vtOouqroGKo1" }).eqJoin("notificationId",...

> What is the final solution for this issue of changefeeds dropping? Not using RethinkDB anymore 🙈 > I am wondering from an engineering standpoint, how this issue was solved...

Wercker failed even though all tests passed 🤔

> Would it be possible to use the `[Symbol.asyncIterator]` protocol so that it's possible to pass the cursor itself directly into the for-await-of loop? Already possible!

In case you don't want to break existing users the `Cursor` could also expose the async iterator under a key, something like this: ```JS const messages = await db.table('messages') .changes({...

Of course, you can get the same thing done today too. This doesn't add any new behavior—the problem is other libraries expecting async iterators as input. For us, that's [GraphQL...

Yeah I'll probably just whip up a quick PR, I just wanted to give @neumino a chance to tell me not to before I go do all that work. :blush:...

See PR #371. I made it a method and called it `cursor.asyncIterator()`, comments welcome!