So high, so low, so many things to know

[mindstalk]

Still job hunting.
Realized that Scala and Clojure are functional languages at least some people pay people to use, I should go learn them! (I have a functional bias.)

As my algorithms studies continue, it's scary to look back and realize how much CS is out there that I didn't even know I didn't know, back when I was working. Both the stuff I learned in grad school (computer theory, OS concepts, graphics, programming language implementation) and the stuff I've learned since (graph algorithms, non-trivial dynamic programming, quickselect, heaps/priority queues, AVL trees...) What did I actually get hired on? Structured programming, lists, trees, recursion, Big-O analysis, hashes (thanks to Perl on a previous job, not any class I took.) Well, you really can do a lot with that. But man.

Comments

[mtbc]

I've certainly found paid uses for graph algorithms, priority queues and suchlike. What interviewers actually ask me about are things like OOP design patterns.

I'll get to do more substantial JavaScript someday (rather than just tweaking web pages with jQuery and whatnot), I'm curious to what extent that does let me write functionally.

I'm really glad that I got to write Haskell for years, it was a real pleasure. Annoyingly, just as we'd got our move to Scotland all sorted, a New Jersey company asked if I could interview with them for Haskell work. Ah well, with luck their actual hire was appreciative.

[mindstalk]

Graphs have come up a lot in interview, assuming I'm not overcomplicating things. So, scary again given that I was clueless about them way back when, also what I know now is entirely from Skiena, not any class. Maybe because I skipped a proper UG degree. Grad "algorithms" course was "analysis of algorithms" and mostly about torturing summations, I barely remember a thing.

Patterns haven't come up much, though one phone screener did ask me about them the other day. I gave a vague answer and had to beg off otherwise.

Yeah, I've heard JS is at least more functional-friendly than Java, or pre-8 Java anyway. Scala seemed close to "ML features with C-ish syntax and people use it", though, and Clojure is basically Lisp.

My one experience with Haskell left me skeptical I'd want to try to use it practically. But that was years ago, I'm told things may have improved.

[mtbc]

Haskell does some neat compile-time optimization but performance can be tricky; I definitely want to try out the new Strict and StrictData pragmas.

I took a masters-level algorithms course at Ohio State: yes, plenty of big-O (again). Though, I don't think it had quite the recurrence relation solving from my undergrad at Cambridge; at one point there I was happy when a friend told me that I could figure the complexity of interdependent functions from the largest eigenvalue of the dependency matrix.