Benjamin Tyner

Yes, it would be good to get some clarity on this. In the meantime our only option is to downgrade back to shiny version 1.5.0 (the most recent version not...

I agree wholeheartedly. Maybe these functions could expose a `bigint` argument which the user could set to `"integer"`, `"numeric"`, `"character"`, or `"integer64"`, with similar behavior to what many `DBI::dbConnect` methods...

In my opinion, normalization of weights is irrelevant here, and so the documentation needs updating. However, we might want to offer another argument to control whether the MLE is computed...

After thinking about this some more, I think the documentation is largely correct, but I would add a clarifying statement along the lines of: > If 'weights' are frequency weights,...

I took a look ... don't you need to add a formal 'method' argument to the function alist? Personally I don't see a need to actually normalize the weights when...

Thanks, it's working now. I also took a stab at correcting the [wikipedia article](http://en.wikipedia.org/wiki/Weighted_arithmetic_mean#Weighted_sample_variance).

I agree that sampling and reliability weights are not equivalent, and I am dubious that normwt=TRUE would ever be appropriate for sampling weights.

To me, the normwt argument is only for specifying the kind of weights. In many cases, sampling weights are realizations of random variables, in which case I don't think they...

That link doesn't work for me, but I see nothing wrong with the existing treatment of sampling weights: ```R set.seed(6860) x

My mistake; was not aware of the distinction between frequency and sampling weights. I must retract my statement about sampling weights being realizations of a random variable; what I had...