Advanced Compilers: Self-Guided Online Course

Languages that are context-sensitive? C, C++, Java, JavaScript.

Examples of tools based on the starting expectation that languages are context-free? Yacc, Bison, Jay.

Examples of the problems this causes? Well we're using the wrong tool for the job, right from the start. Instead of using an appropriate tool the first thing we do is bend our tools out of shape. We use side-effect actions to subvert the tool's model in an uncontrolled way. We don't get the full benefits of the tool's original model and can't rely on its guarantees.

I teach compilers at a big university. And I would love to hire graduates with compiler skills for my startup, but find it difficult.

There are several structural problems that conspire to keep students from acquiring knowledge in low-level programming domain like compilation: a course needs to fit with the rest of the curriculum, and with student interest. I cannot get students interested in lexing and parsing, because they hear all day how exciting deep learning and big data are. Lexing and parsing are deemed solved in the 1960s, a sentiment I disagree with. In addition, classical theoretical computer science (e.g. automata and languages) is rarely taught with enough depths in many universities, so students lack the technical background. You can cover this in a compilers course (I do) but it takes time, and the students see it as an ordeal. Compilers as a subject is not helped by the most widely recommended book being the Dragon Book which is dated and way too long to be useful for beginners. Compare the Dragon Book with the material available as introduction to machine learning ... Many of the existing textbooks are also not covering modern compilation themes, in particular JITs, and compilation for multi-core and GPUs. I'd say there is currently no good undergraduate textbook on compilers. I could probably, with a lot of work, cobble something reasonable together from my lecture notes, but I don't believe in books, I'd like to do a good MOOC, but building suitable software infrastructure requires more work than I am currently willing to put in.

My department tried hard to remove the compilers course, as "no longer relevant", and "too hard". I threatened to resign if it was not kept as a mandatory course. For now this worked.

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 hacked-up DSLs!) have poor error recovery unless you put in the work. Most LR parsers are also awful by default.

When I was in university most focus was on LL and LR parsers with no discussion of error recovery and more focus on memory usage/bounds than speed. I also have no idea how common it is to teach combinator-based parser grammers these days; stuff like ANTLR and yacc dominated during my studies. This would add another level of unfamiliarity for students going to work on a "real compiler".