gistfile1.md

[Something I learned later: they have major coding test(s) on paper. Biased against otherwise successful programmers with dyslexia.]

Hi [redacted]!

Some thoughts based on my weak understanding of your situation. Please skip anything unhelpful!

You're welcome to email me: tjg@pentaside.org

Initial thoughts

I'd personally consider dropping the class [CS 2110]:

• Zero respect for your time: "They have no problem losing a third of the students (or more)."

• Java+Eclipse are extraneous overhead — and poor user-interfaces to computation. Python's the obvious choice — they teach it, supports interactive programming, and lets you only write code that matters (instead of boilerplate).

• Instructors admit "60% of students have not seen Java prior to taking this course", and are too lazy/incompetent to support additional options like Python.

• Why mandate a one-size-fits-all 1000-page book instead of simply focusing on your programs? (Or at least why not point you to "Design and Analysis of Computer Algorithms" by Aho/Hopcroft/Ullman and "Programming Pearls" by Bentley?)

• The algorithms/datastructures covered by the course are usually available as (optimized, debugged) libraries or StackOverflow snippets.

• They say, "Did you have binary search in your previous course? If so, this should be easy." Cruel! Binary search is notoriously hard to get right — Java had a bug in it for 9 years; professional programmers screw it up all the time.

(Nothing's easy, except falling for the Dunning-Krueger effect. Ironically, Dunning & Krueger developed it at Cornell...)

Techniques that might speed you up

You mention Java (not the concepts) is the main problem. Techniques to consider & combine:

• "Pseudocode programming process": a) type pseudocode, b) turn it into comments in your sourcecode file, c) implement it between the comments.

• First work in a programming language you like. (Python?) Then translate your debugged, polished, tested code into Java.

• "Test-driven design": Write unit-tests first. (Don't forget edge-cases.) Then write the code which makes your tests pass.

• Do your work on paper, then implement the code. Because typing in symbols is an impoverished user-interface. (Paper's also weak, but more flexible than just symbols.)

• Hammock-driven development: Lie back & think. [Video, transcript]

• Use lots of print statements to visualize what your code really is doing.

These techniques attack a central problem of Java — juggling multiple ideas simultaneously. (Declaring types, public/private...) Mental focus is vital.

Another problem with Java is people constantly use custom interfaces instead of generic data. Example:

// No preexisting code in the world supports your new Point type.
// Java devs need heavyweight IDEs to navigate the rat's nest of custom interfaces like this.
new Point(2,5);

// Zillions of functions operate on Map types.
{x: 2, y: 5}

Institutional critiques

Helps me act effectively, when I learn why an institution is so dysfunctional.

• Disciplined Minds (Schmidt)

• Constituent Imagination (Shukaitis, Graeber): "For the most part, universities were dominated by figures who were scholars but in no sense intellectuals."