larsberg

21 ms·

151.
by larsberg 13y ago
Certainly! SML/NJ was a research platform first and high-performance/implementation second. Defunctorization helps, but from talking with Matthew Fluet, most of the perf comes from the combination of monomorphization and whole-program compi
152.
by larsberg 13y ago
You will need to speak Mandarin. I was in Shenzhen (for an academic conference) recently and hit the electronics stores, all of which are staffed by folks who pretty much exclusively speak Mandarin. That said, it was awesome just to wander
153.
by larsberg 13y ago
It will be a challenge to beat SML/NJ if the implementation has not progressed much since then, much less MLton. Since the 90s, SML/NJ has gotten a new backend, garbage collector, and has gone through a significant amount of performance tun
154.
by larsberg 13y ago
One of the former chairs of our department (CS at the University of Chicago) has a saying, "all curriculum reforms initially succeed." So I'm not surprised that the pilot program had great pass rates, as pilot programs are usually staffed b
155.
by larsberg 13y ago
That's an intro course for non-majors. Scheme is use in the curriculum for majors (the 151 course), but it uses the How to Design Programs text, not SICP. The Honors curriculum uses Haskell as its functional language, with touches on some o
156.
by larsberg 13y ago
Two things that make me suspect this information is anecdotal rather than from a hiring manager (though I could be wrong): 1) I left MSFT around 2006, but the offers I made to (admittedly, mostly blue chip) candidates in 2002-2004 were clos
157.
by larsberg 13y ago
tl;dr The easiest way to paint yourself into a corner on a technical interview question - especially a trick one - is to answer it without asking clarifying questions about boundary conditions, usage patterns, data set sizes, target hardwar
158.
by larsberg 13y ago
Do you think everyone would take those random days off on a regular basis if your founders instead worked 6 days a week and never took vacations? (honest question based on your situation, not rhetorical/sarcastic) Sometimes, particularly wi
159.
by larsberg 13y ago
People mimic their management chain. If the CEO/managers don't take vacations, neither will their directs. One of the most balanced (and productive) teams I've worked on was under a general manager who had 8 kids and made it a point to atte
160.
by larsberg 13y ago
I completely agree. I've worked at a company where we had explicit weeks taken out of the schedule to do "app building" where we were expected to do a mix of kicking the tires on the product and experimenting with random tech we wanted to p
161.
by larsberg 13y ago
I'm not sure I agree completely with his characterization of what happens. I've noticed that people, particularly in larger organizatons, will start shoehorning their "hacking work" into real projects if it is left out for too long. So, you
162.
by larsberg 14y ago
Too bad you're getting downvoted! The glassdoor stats are useful, but as you said seem to reflect disgruntled or people on the way out. For example, the pay ranges were at the bottom of the level band (which was tied to the job titles) corr
163.
by larsberg 14y ago
Most of the time in whole-program compilers such at Manticore and MLton (> 60%, though it depends on the particular program) is spent in the combination of parsing the input source files and running GCC over the generated assembly to pr
164.
by larsberg 14y ago
You're missing one of my favorites, "Involving undergraduates in research." Especially for top students, I feel this element is key (I'm biased). Even for undergrads with no desire to go on to do academic work, it's a good experience to wor
165.
by larsberg 14y ago
It's still a research language today. Honestly, we would need to make a fairly major push --- which would freeze all resarch work and publications --- to get to a point where it is ready for general use. We're working on some grant proposal
166.
by larsberg 14y ago
Certainly, that's true! As I mentioned in my other far-too-wordy response, though, if you just CPS convert and run a program, it will be far too slow (5x the funcalls != 5x the fun...). So, it's the combination of CPS conversion and later o
167.
by larsberg 14y ago
You can keep the original tokens. The challenge with debugging such programs is really the integration with the rest of the optimizations. CPS transformation turns your program into a ton of tiny functions. Then, any reasonable optimizer wi
168.
by larsberg 14y ago
If I understand it correctly, the Rust issue is that tail calls are expensive in the context of LLVM and the calling convention that they are using. MLton (like Manticore, which I work on) uses a different code generation backend and callin
169.
by larsberg 14y ago
By 0.9 offices / headcount I mean the ratio of space allocated to your team (number of physical offices) versus the number of people. The closer that is to one, the more people get their own, private office. At least when I was at Microsoft
170.
by larsberg 14y ago
> everyone wants to be in charge Seriously? When I was at Microsoft, it was like "hot potato" to have to give up a coveted dev role and take on dev lead responsibilities. In particular, being a first-level manager was treated as a sort
171.
by larsberg 14y ago
I don't know about google - all the PL people I know who went to google no longer work on PL stuff, though I don't know if that's because of working at offices where PL work is not supported rather than a lack of company-wide freedom. From
172.
by larsberg 14y ago
#1 is also pretty achievable. For example, if you get a Ph.D. studying how to implement compilers and virtual machines, you basically have your choice of which place that implements compiler or VMs to work at (source: I am about to defend m
173.
by larsberg 14y ago
Strength coaches call this the "novice effect." It's why you can see an untrained person's squat increase if they ride the exercise bike a few times a week. If your goal is optimality, the question is: is your training program the most effe
174.
by larsberg 14y ago
... and all of the Suggested Posts.
175.
by larsberg 14y ago
This challenge covers much of compilers research. Many heuristics and techniques that are used in practice in compilers (certainly in ours and GHC and ocaml and other research ones I've talked with) have solutions we've all verbally shared
176.
by larsberg 14y ago
We believe so! Our project leader is focused on how we can parallelize general-purpose applications easily. With more and more people writing in static functional languages that have relatively poor parallel scalability without massive pr
177.
by larsberg 14y ago
I completely agree. As someone who works on a parallel functional language, it's very hard to sell a parallel language that isn't as fast as parallel fortran or hand-tuned C code that uses pthreads and the fastest parallel implementation of
178.
by larsberg 14y ago
It would be neat to see outside the USA. When I was just out of country for some conferences, my time spent in HK and mainland China were dominated by Android phones (especially that monsterous Samsung one). Literally, I saw more Windows Ph
179.
by larsberg 14y ago
Have you tried using any of them? Say what you will about Google Reader, but at least it didn't crash (most of the standalone apps) or require some sort of strange interpretive dance with your fingers to browse your articles and mark them a
180.
by larsberg 14y ago
Completely agree with you - I frequently used to ask experienced candidates questions of the form, "tell me about one very difficult bug or performance issue encountered during your two years working on X." Unfortunately, the number of flat

More ›