Sridhar Ramesh

We must destroy time's arrow, so that in the future, we may return to the past.

I hope Donald Trump dies. Moderators: I hope Donald Trump dies.

I find this slapdash reporting implausible. It's hard to believe the president would knowingly target young girls.

Americans are posting in paranoid hysterics, as though they're in any danger, as though any reaction from Iran would reach over here. Relax. As I understand from the theory of war and its endgames, we will be safe, as the Ayatollah can only attack diagonally.

Nothing good ever happens. Bring back nothing ever happens.

I regret nothing. Specifically, the nothing I've accomplished with my life so far.

I understand how a non-racist could know of the N word. They could have read it in classic text "The Adventures of Huckleberry Finn". But what's damning is knowing how to pronounce it. That goes beyond mere academic well-readness. Such knowledge can only be learnt on the street.

I understand how a non-racist could know of the N word. They could have read it in classic text "The Adventures of Huckleberry Finn". But what's damning is knowing how to pronounce it. That goes beyond mere academic well-readness. Such knowledge can only be learnt on the street.

When I heard what that guy said at the BAFTAs, it turned my stomach. I hate N words. I hate hearing N words and I hate seeing N words.

It is important to remember that, despite popular portrayal, only about 10% of people with Tourette syndrome suffer from coprolalia. The other 90% just shout "the N word" and not the actual word.

It is important to remember that, despite popular portrayal, only about 10% of people with Tourette syndrome suffer from coprolalia. The other 90% are genuinely racist.

A lot of ignorant commentary about the outburst at the BAFTAs, from people who haven't bothered to do five seconds of research. In Britain, that word means "cigarette", and it is spelt with the R before the E. It is customarily pronounced with the "hard R" in Scotland, but non-rhotically in England.

There is only one way to make amends. Michael B. Jordan and Delroy Lindo must now be permitted to call one person of their choice the N word (once each). Horrifying as such Hammurabian justice may seem, they must be given a kind of "N word pass".

There is only one way to make amends. Michael B. Jordan and Delroy Lindo must now be permitted to call John Davidson the N word (once each). Horrifying as such Hammurabian justice may seem, they must be given a kind of "N word pass".

I aspire to someday write a joke that is used on an officially televised or streamed comedic program. Just one validated joke once would suffice.

I don't know the details of the dplyr implementation of mutate but I note it does not have this issue.

Even of R, which has been my least favorite language to use for what I think of as programming, I do appreciate the particular usability Colin has been pointing out for data exploration tasks.

Different languages for different purposes often makes sense.

In a professional sense, a few years ago, I had a surprisingly good time using Kotlin for work.

In my life history, I cut my teeth on C and, while it's not now what I would reach for for most tasks, I think it's very understandable why it became so dominant and influential for so long.

But I find things to like about many rather non-Haskelly languages too. Continuing in an academic wankery vein, I've enjoyed much about Scheme and Prolog, despite also finding other things about their largely type-less design very frustrating.

Also in a similar statically typed functional programming vein, I wish Haskell had adopted the module system of ML (as in OCaml or Standard ML), rather than its type class system which I rather dislike (contra most Haskell aficionados).

Other languages in a similar vein to Haskell that I am interested in, but with dependent typing, are Agda and Idris and such. Typical PL/type theory stuff. Though those are not nearly ready to be used (or even intended to be used) in such a general way as Haskell or more mainstream languages, yet.

I find Haskell to be full of warts that frustrate me, to be clear. My ideal language would be very different from Haskell. But among languages that currently exist, it does have many traits that many other languages lack that I find often useful.

Incidentally, I don't think this does quite what you want it to do. Consider:

myMutate <- function(data, new_column) {
data$new <- eval(substitute(new_column), data)
data
}

newval <- "wrong"
addVal <- function(data, newval) {myMutate(data, newval)}
addVal(data.frame(x = 0), "right")

Yes, it's the first-class environments. Which is not just the combination of lexical scoping and lazy evaluation, which you had originally mentioned. Lexical scoping with lazy evaluation does not suffice. The ability to request evaluation in a different environment is a whole extra feature.

There should be a way for me to post these posts that are slightly longer than the character count constraint but still just a few reasonable sentences without having to resort to breaking them in awkward places or taking screenshots.

Which is opposite to the behavior that Colin is pointing out his fondness for in `mutate`, etc. The behavior Colin is fond of is the ability to disable lexical scoping and re-achieve dynamic scoping, via `eval` and `substitute`. It's not R's default lexical scoping that achieves this.