« first day (1716 days earlier)      last day (1042 days later) » 

00:04
The plagiarism question is pretty hot.
00:39
Say, how do you move a comments thread to this classroom? Do I not have enough karma, eh, reputation, for doing that?
@VictorEijkhout It's only a moderator ability, but we can talk here if you prefer!
I think my most recent comment is basically all I can say on the matter, tbh. I'm no expert in compilers or in FP, so I can basically only parrot back what I've heard. @Rusi has pretty deep FP expertise, and I would trust anything he says on the matter a million times more than anything you hear from me 🤣
Heck, my graduate degrees are even in music, not in CS. My career has just taken rather unexpected turns over the years.
I do have an undergraduate degree in CS from ages ago, but at this point, most of what I know I've learned on the job teaching CS.
@Buffy That's slightly crazy to me since it only got two answers.
@BenI. I'm open to hearing about actual cases of FP codes showing good performance. Of course we'd have to demarcate what the area of investigation is. Here is an example of something I investigated and where I get 50+ times speedup on 56 cores

https://victoreijkhout.github.io/pcse/omp-examples.html#N-bodyproblems
@BenI. Just curious, what do/did you do in music? Music was always a hobby for me, but a pretty serious one.
00:54
@VictorEijkhout I'd not be a great primary source. My google-fu is pretty good, but I imagine yours is as well and you can dig up the same stuff I can.
@VictorEijkhout I am a conductor!
But these days, I mostly teach computer science to very advanced high school students.
@BenI. Nice. Orchestras, choirs? That's one thing I never did. But then I'm mostly into chamber music and small ensembles. No conductor required there.
Mostly choirs, the lately. I've been doing a lot of band work and occasional orchestral stuff
*though
01:25
@BenI. Ok, I found this: benchmarksgame-team.pages.debian.net/benchmarksgame Haskell is typically 3x slower than C. Given that where I live people get upset about a 10-20 percent loss, that pretty much disqualifies Haskell.
Also I have a serious beef with the choice of problems. For instance, I looked at the N-body code, which is !!!4 planets!!!! That's a joke. For heaven's sake, do this code on a globular cluster of a billion stars. I'd be surprised if Haskel then came within a factor of 30 of C.
01:46
How does that compare to bytecode languages?
 
1 hour later…
02:49
@VictorEijkhout
For the Haskeller getting Haskell down from 100-fold slower-than-C to 3-fold slower is understandably a win
For the performance-maven 3-fold slower than C throws it out the water
Both have their place.

I am with Ben here in saying "I dont know much here..."
I would have thought top-dog(s) here would be C,Fortran not C,C++...
But what do I know

Why someone like you should explore something like Haskell — if at all — is to explore expressivity of algorithms not their final performance.
For performance+FP there are 2-3 things to consider
[and once again remember, I am not in my element here]

1.
There are performance-oriented FPLs nowadays — Julia and Rust come to mind.
Not exactly FP by Haskell standards but somewhat in the mid-zone

- In C assignable is default, 'const' is exlicitly specified
- In Rust immutable is default 'mut' is explicitly specified

You may think it small but when Rust flips C's DEFAULTS its a sea-change in programmer attitudes.

Your question will of course be:
03:23
@ScottRowe
I like to think of insight as an asymptotic possibility

Speaking of (with) music

Here is my expanded version of the quadruple:
Para-Pashyanti-Madhyama-Vaikhari

Idiomatic English rendering
verbiage-mentalese-insight/intuition-transcendence

However the literal sanskrit says it better than the English

The Greate Other - Seeing - Intermediate - Words

And further elaborated (in a LOT of words)
https://philosophy.stackexchange.com/a/64818/37256

Speaking of music....

[ @BenI. Being a conductor was always something I thought impossible, unimaginable. But in school days I serious
So speaking as a teacher:

I think Level 0 — words (vaikhari) — matter as a serialization protocol
If my words are ungrammarly, ill-spelt etc all the higher levels are closed off to the listener

So words are necessary but not sufficient

Concepts — mentalese (madhyama) — is where the majority of our action lies
But its important to remember it too is necessary, not sufficient:

Students may understand everything and yet... something completely missing

So its important to keep aiming for insight/intuition/ "feel"
03:44
@Rusi I was responding to a linked SE post that was wondering "why is haskell so fast", so clearly some FPers are concerned about performance. But yes, my first interest is in expressivity. Hence my original question what sort of assignments you'd give in an FP oriented programming course.
@VictorEijkhout you could say Haskell is fast by the standards of toy academic languages.
Re your question I would be pleased to continue!
Ive asked you ( @VictorEijkhout ) there if some rewording could help span some communication gap. Pls see.
Of course we could continue along a different vector if that helps more
Re expressivity: As a teacher I'm a "Lo level guy": if students say "char *" or they say "string" it's a big shift (in my book)
In other books it's probably twiddle dee Vs twiddle dum!
 
16 hours later…
20:08
I know someone who works with a Teradata brand machine. Size oafr
Oops - size of a fridge, super fast for manipulating large amounts of data. 30+ years ago, I saw a machine, size of a small fridge, that did realtime modelling. The demo was a flag flying, in full detail and color. What language? No idea. But "Speed is just a question of money. How fast can you go?"
@Rusi I would say that knowledge is asymptomatic. My sense of insight is that it just hits you, and it is inarguable, but also can't be conveyed. So, not really 'useful', but gives perspective.
I LOVE the statement from one of Frank Herbert's books: "Seeing moves." If you grok that, it's insight.
Another great example is the movie "Arrival". If you get it, you get it. If not, not.
20:56
Gosh, I meant 'asymptotic' above. Too late to edit now...

« first day (1716 days earlier)      last day (1042 days later) »