Ozan Adıgüzel

I don't have much experience either but here's my two cents. I personally like the **squash and merge** strategy, which condenses PR's commits into a single commit on the main...

There is a potential size limitation that would cause an issue with large number of variables to match. ``` ?regex ``` > Long regular expression patterns may or may not...

Following up this issue, I can't create a new instance either: ```r vm = gce_vm( name = "", username = "", password = "", ) # Error in value[[3L]](cond) :...

Thank you for taking the time to work on this. Here is my attempt to use `gce_make_network()`: ```r ## works fine when starting a VM that was not running vm...

```r options(googleAuthR.verbose = 2) ## works fine when starting a VM that was not running vm = gce_vm( name = "my-existing-vm", network = gce_make_network( network = "my-network", subnetwork = "my-subnetwork"...

This small [PR](https://github.com/stan-dev/loo/pull/188) presents a suggestion to extend the existing `gpdfit()` function to also return the marginal posterior density of `k` using the code @avehtari posted above with a few...

Sounds good, I added this normalization term estimation with the trapezoidal rule to the PR thread along with a couple of implementation questions.

@avehtari - I can update the function with normalization term estimation using the trapezoidal rule once you decide what to return. Below are some options: ```r # normalization term estimation...

I noticed a small side effect of [this recent commit](https://github.com/stan-dev/loo/commit/bffab8cd46270cba8e7dd1d09c12b880dc8dda2c). That is, when the estimation doesn't work, k-hat and sigma-hat are not necessarily `Inf` and `NaN`, respectively: ```r set.seed(147) x

- tried to update `gpdfit()` to reflect the improvements @avehtari mentioned above - implemented the full suite of `dgpd/pgpd/qgpd/rgpd()` functions - cleaned up and condensed some internal helpers @jgabry; the...