ML Family Workshop
syslog.cl.cam.ac.uk
syslog.cl.cam.ac.uk
links to papers: http://okmij.org/ftp/ML/ML14.html
I seldom hear about the others except on research papers.
I see one of the abstracts talks about a "classless society" as a pun relating to object orientation. This makes me groan a bit, as there seems to be a general tendency in FP to make superficial quips, puns or observations like this while the community of FP programmers remains totally divorced from reality.
A good example is the notorious video interview https://www.youtube.com/watch?v=iSmkqocn0oQ in which Simon Peyton Jones laments the fact that pure functional programming languages just make the CPU heat up and nothing else (i.e. they have no useful side effects.) He probably wasn't being completely ingenuous but it comes across as a lack of self-awareness: FP is all about calculation and expressions, the program is an input and the value it evaluates to is the output... That's all you can expect. You need some other paradigm to "do" things, IO monads notwithstanding.
Conan Elliot has a fun blog post on the c preprocessor which dances around this issue a bit: http://conal.net/blog/posts/the-c-language-is-purely-functio...
You're not the first person I've seen that's misunderstood this video, I can only guess at why... One thing it's probably worth remembering is that Haskell has been a research platform for many years, and some of the Haskell features that people take for granted today did not exist in earlier versions.
SPJ is not saying that Haskell is useless now, but it was close to useless for creating programs for a number of years. As a case in point, try to imagine Haskell without monads. Hopefully knowing that Haskell did not have monads at one point helps you understand the historical context of this discussion.
In any case, I don't think I have misunderstood: I just don't agree with his notion that a hypothetical purely functional language is "useless", as I explained.
I reckon a lot of people want to use that kind of machine, though it rules out systems programming. See for example the requirements expressed in this article: http://www.americanscientist.org/issues/pub/calculemus
No, I'm not saying that. Listen to the video...
(3:34): Nevertheless, we put up with that embarrasment for many years until we figured out how to combine, in a single language, a single programming (...), we can combine effectful computations with effect-free ones, without making them pollute each other, right. The type system keeps them apart.
(...) Can't quite make out this word, "methodology" perhaps?
In any case, as you can hopefully see from reading that quote, SPJ is talking about the situation with Haskell in the past. Until they worked out how to do effectful computation in a pure functional language Haskell was basically useless, after they worked it out then Haskell was not useless. Do you understand what SPJ was saying now?
But in any case that's not my point. I am arguing that Haskell was never useless in the first place.
(You seem to be ignoring the substance of my comments, so continuing this thread is probably a waste of our time.)
Okay, let's get break this down. The Haskell project was started in 1987. Version 1.0 of Haskell was released in 1990. Version 1.0 had no monads (they didn't come until version 1.3). If you had to code something that was both pure functional and useful in version 1.0, could you do it?
"he is saying they are moving in that direction..."
Yes, of course, because "useful" is a relative term. It's a goal, but the work on making a language more "useful" is never complete. The sketch is not important here, it's just a simple sketch, listen to SPJ's words.
For me the context here is that Cambridge is a massively elitist centre of frivolous intellectual pursuits, in a country with the highest levels of inequality in Europe. http://www.theguardian.com/society/2014/aug/28/closed-shop-d...
Perhaps I have a chip on my shoulder about this, but I actually did my undergraduate CS degree in Cambridge before I knew anything about the world so I am at least somewhat qualified to comment on that milieu, from a political point of view. It's cosy but relevance to the everyday life of the masses is not a strong point. I don't have any comment on whether JavaScript implementation details have political consequences.
And is your claim really that people in the functional programming community shouldn't make jokes in their talks until FP lives up to your notion of practicality? How many banks need to use Haskell, how many switches need to run on Erlang, etc etc, until you're satisfied?