Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Advanced Compilers: Self-Guided Online Course

cs.cornell.edu

Post reply on HN
Print view View original

191–200 of 241 posts

Re: Advanced Compilers: Self-Guided Online Course

#191

Post by seanmcdirmid »

ibains wrote: post #182
seanmcdirmid wrote: post #81

Earlier quoted context omitted.

You are thinking about it the wrong way. In this niche, you will almost never find people with the exact skills you are looking for. Better to find people with tangentially related skills and experience and let them grow into the role. Academia is never going to be comprehensive enough to prepare students for every niche that has some demand. Rather, they should just focus on teaching a core that will help them learn…

packrat parser combinators are recursive descent with infinite look ahead. That’s the problem, engineers are under educated

The problem is some engineers are ideological, you don’t want to work with those. If you have this argument during an interview and the interviewing engineer is insisted that X is always the right choice, run away.

Anytime a commercial language gets bogged down in parsing, unless that is core to the product (and mostly it isn’t), also run away. Parsing represents like 1% of the work that needs to be done on a compiler, even if it’s an IDE parser that needs to be really incremental (the only worse thing is to get stuck on lexing).

Re: Advanced Compilers: Self-Guided Online Course

#192

Post by throwaway_pdp09 »

Kranar wrote: post #190
kingaillas wrote: post #170

Earlier quoted context omitted.

You realize the poster you are responding to is the creator of D? And has spend his career writing compilers? I'm not sure he needs to have someone explain what a context free grammar is.

I understand you wish to flatter WalterBright, but please note that your comment is fairly toxic and not well suited to Hacker News. Please stay on topic and feel free to contribute to the points being made. This kind of hero worship often just degrades discussion and is better left to other forms of social media.

Yours is the toxic comment. Appeal to authority is fine if the authority is an authority and WB has proven his. Suggestions of intention to flatter and accusations of hero worship are pretty gross. @kingaillas's point is entirely valid IMO.

Re: Advanced Compilers: Self-Guided Online Course

#193

Post by throwaway_pdp09 »

lifthrasiir wrote: post #189
onei wrote: post #152

Earlier quoted context omitted.

Isn't that pretty standard for a parser? When was the last time a compiler bailed on the very first error it hit and refused to do anything else? The solution is to pick synchronisation points to start parsing again, i.e. ; at the end of a statement or } at the end of a block.

How can you be sure that that } is the end of a certain defined block? This most importantly affects the scoping and in many cases it's ambiguous. IDEs do have rich metadata besides from the source code but then parsers should be aware of them.

You're ignoring the ; which are sync points.

> How can you be sure that that } is the end of a certain defined block

If it's not in a string, what else is it but a typo? If a typo, it fails to parse but so long as it doesn't crash, fine.

Re: Advanced Compilers: Self-Guided Online Course

#194

Post by throwaway_pdp09 »

IshKebab wrote: post #134
throwaway_pdp09 wrote: post #125

Earlier quoted context omitted.

Does that issue go away if you use packrat vs some other means?

Yes. Go and look up Tree Sitter and its references.

Parsing (use of rather than theory) matters as it affects my work. So I followed up.

See https://youtu.be/Jes3bD6P0To

Tree sitter is based on LR parsing (see 23:30 in above video) extended to GLR parsing (see 38:30).

I've had enough of fools on HN posting unverified crap to make themselves feel cool and knowledgeable (and don't kid yourself that you helped me find the right answer by posting the wrong one). Time I withdrew. Goodbye HN.

Re: Advanced Compilers: Self-Guided Online Course

#195

Post by lifthrasiir »

throwaway_pdp09 wrote: post #193
lifthrasiir wrote: post #189

Earlier quoted context omitted.

How can you be sure that that } is the end of a certain defined block? This most importantly affects the scoping and in many cases it's ambiguous. IDEs do have rich metadata besides from the source code but then parsers should be aware of them.

You're ignoring the ; which are sync points. > How can you be sure that that } is the end of a certain defined block If it's not in a string, what else is it but a typo? If a typo, it fails to parse but so long as it doesn't crash, fine.

Maybe my wording is not accurate, imagine the following (not necessarily idiomatic) C code:

    int main() {
        int x;
        {
            int x[];
        //
This code doesn't compile, so the IDE tries to produce a partial AST. A naive approach will result in the first } matching with the second {, so `x += 42;` will cause a type error. But as noticable from the indentation, it is more believable that there was or will be } matching with the second { at the caret position and `x += 42;` refers to the outer scope.

Yes, of course parsers can account for the indentation in this case. But more generally this kind of parsing is sensitive to a series of edit sequences, not just the current code. This makes incremental parsing a much different problem from ordinary parsing, and also is likely why ibains and folks use packrat parsing (which can be easily made incremental).

Re: Advanced Compilers: Self-Guided Online Course

#196

Post by the-smug-one »

WalterBright wrote: post #158
cinericius wrote: post #69

Earlier quoted context omitted.

Do you think it is the responsibility of academia to teach "industry relevant stuff"? I agree that courses generally don't teach judgement well, but I do think that it is the workplace/industry's responsbility to train their engineers and not expect fresh grads to be fully equipped to work on something that specific.

> Do you think it is the responsibility of academia to teach "industry relevant stuff"? In college I learned a lot of math and how to do things like calculate stress and bending. I learned nothing about material finishes, material selection, fastener selection, tolerances, drafting, etc. But the latter was easy to pick up on the job, while learning mathematical analysis on the job just doesn't happen.

And learning PHP, React, and MongoDB is easy to do on the job.

However, I think that there are industry practices which may inform what students are taught. For example, process algebra is rarely (if ever?) used in industry. There are other formalisms of comparable difficulty and academic interest which are used however, perhaps teach those instead?

Re: Advanced Compilers: Self-Guided Online Course

#197

Post by morelisp »

haberman wrote: post #168
morelisp wrote: post #128

Earlier quoted context omitted.

Packrat parsers are notably faster than recursive descent parsers (also critical for IDE use) and by turning them "inside out" (replacing their memoization with dynamic programming) you get a pika parser which has very good recovery ability. There are varying techniques to improve error recovery for all forms of parsing but hacked-up recursive descent (certainly the most common kind of parser I still write for my hac…

> Packrat parsers are notably faster than recursive descent parsers I think this needs to be qualified. I don't think a packrat parser is going to beat a deterministic top-down parser. Maybe the packrat parser will win if the recursive descent parser is backtracking.

True - the performance of a top-down parser is going to depend on if/how often it backtracks and how much lookahead it requires. But this requires control over your grammar which you might not have i.e. you are parsing a standardized programming language. Practically speaking, unless you want two separate parsing engines in your IDE, that Lisps are actually LL(1) and Python and Java are "almost LL(1)" doesn't get you closer to parsing C++.

Packrat parsers are equivalent to arbitrary lookahead in either LL or LR grammars.

Re: Advanced Compilers: Self-Guided Online Course

#198

Post by petr_tik »

weakfish wrote: post #180
petr_tik wrote: post #166

Earlier quoted context omitted.

As i mentioned in another thread, a uni course that has students work on an existing codebase instead of writing something from scratch might be even better - though harder to grade. Imagine if one could do a major feature contribution to clangd/pyls/rust-analyzer as part of their advanced compilers course. The student gets a better understanding of (open-source) software development practices, gains a deeper profici…

My uni, NC State, has a course (CSC 326) called Software Engineering that has students contribute to an existing project created by the staff in teams, with the best contributions being merged into master for the next semester students to work on.

That sounds pretty cool. Do you get a choice of what kind of project you want to contribute to?

Re: Advanced Compilers: Self-Guided Online Course

#199

Post by chrisseaton »

ibains wrote: post #174
chrisseaton wrote: post #95

Earlier quoted context omitted.

> Good luck hiring an engineer with advanced compiler courses knowing any of this. Have you tried raising the level of compensation you offer? There are engineers elsewhere doing this at other companies you could presumably get if you put your hands in your pockets?

We pay very very well, all my peers from NVIDIA are above 700k. Why would you presume anything about salary?

It's not true that there aren't engineers with these skills - there are thousands of them in the US and more worldwide - so if you can't hire them then either:

* you aren't looking (seems unlikely)

* there's something toxic about your team that means they'd never work for you and you should fix that (hopefully unlikely)

* you aren't paying enough to make the move worthwhile for them (seems likely as people never want to pay the going rate)

That's why I presumed salary. Double your salary and you may suddenly find you're being flooded with contacts from the compiler engineers at places like Google. If that's the case then your only problem was the compensation you were offering and not a problem with education.

Re: Advanced Compilers: Self-Guided Online Course

#200

Post by daniel-levin »

ibains wrote: post #176
vector_spaces wrote: post #96

Earlier quoted context omitted.

I saw this great interview recently with Anders Hejlsberg at MSFT on how modern compiler construction differs from what is taught in traditional University courses. Is this what you're alluding to? After watching that interview, it's strange to read a comment like yours. While the architecture is completely different, it doesn't frankly seem like that big a leap to go from "senior engineer with X years working on thi…

Yes, more validation - compilers as taught in universities need improvement. So, a startup of 10 people should hire and train people for months. Raise a $2M seed round, have 12-18 months to show ProductMarket fit and as a founder teach all engineers their jobs? How many engineers go into compilers after school? So university education is all they know. The industry is built on qualified engineers. NVIDIA ceo was ahea…

>> Why are compilers ones behind the curve?

I suspect it’s because there aren’t a lot of high quality libraries you can integrate into the backend of compiler tools that don’t run into license issues pretty fast. Imagine if GNU binutils was more permissively licensed and as modular as clang? Then developing novel, non-GPL’d compiler infrastructure could depend on BFD - the boring part that working on won’t bring bonafide improvements to your new compiler. Another factor is that LLVM’s quality and ubiquity has reduced the monetary and technical upside to pursuing new opportunities in compiler development.

Post reply on HN