Exercise docs refer to another exercise that might not be implemented

BNAndras · February 10, 2025, 3:41am

exercism/problem-specifications/blob/f8c8c35f4553d55419c5ac9fce54f08f0d93cefc/exercises/simple-cipher/description.md?plain=1#L60


      
          
          The weakest link in any cipher is the human being.
          Let's make your substitution cipher a little more fault tolerant by providing a source of randomness and ensuring that the key contains only lowercase letters.
          
          If someone doesn't submit a key at all, generate a truly random key of at least 100 lowercase characters in length.
          
          ## Extensions
          
          Shift ciphers work by making the text slightly odd, but are vulnerable to frequency analysis.
          Substitution ciphers help that, but are still very vulnerable when the key is short or if spaces are preserved.
          Later on you'll see one solution to this problem in the exercise "crypto-square".
          
          If you want to go farther in this field, the questions begin to be about how we can exchange keys in a secure way.
          Take a look at [Diffie-Hellman on Wikipedia][dh] for one of the first implementations of this scheme.
          
          [img-caesar-cipher]: https://upload.wikimedia.org/wikipedia/commons/thumb/4/4a/Caesar_cipher_left_shift_of_3.svg/320px-Caesar_cipher_left_shift_of_3.svg.png
          [dh]: https://en.wikipedia.org/wiki/Diffie%E2%80%93Hellman_key_exchange

This line only makes sense if a track has also implemented Crypto Square. If it hasn’t, this line is misleading. Are there other practice exercises that directly refer to another one? My problem is with another exercise being named, not an exercise touching on concepts related to another exercise.

kotp · February 10, 2025, 5:58am

The immediate thought for practice exercises that (may) reference others is the Amusement Park and Amusement Park: Improved that is on the Ruby track. If the exercise does not, itself, I often do.

Another set of exercises that may is the Resistor series of exercises.

BNAndras · February 10, 2025, 7:14am

Resistor Color isn’t referenced by Resistor Color Duo or Resistor Color Trio. Resistor Color Duo is referenced by name in Resistor Color Trio. I did a cursory check, and all 35 tracks implementing Resistor Color Trio also implement Resistor Color Duo. The three exercises form a series, and it seems odd to me that a track would only choose to implement the third exercise in that series. That said, it’s relevant.

In Resistor Color Duo you decoded the first two colors. For instance: orange-orange got the main value 33. The third color stands for how many zeros need to be added to the main value. The main value plus the zeros gives us a value in ohms. For the exercise it doesn’t matter what ohms really are.

If we don’t mention Resistor Color Duo, we probably should instead add an explanatory sentence describing how orange orange maps to the value 33.

keiraville · February 10, 2025, 7:33am

Of the 75 active tracks:

only elm and pharo-smalltalk have simple-cipher but not crypto-square
only bash and pharo-smalltalk have resistor-color-duo but not resistor-color

It might be simplest for those tracks to add the exercises
.

tasx · February 10, 2025, 12:53pm

The description is overly long for a math-focused problem, with too much non-essential detail. It needs significant simplification, and the extensions section should be left out. Based on my experience, many people might lose interest or feel confused and give up.

Also, I’m not a fan of exercises referencing other exercises. Resistor Color Trio assumes familiarity with Resistor Color Duo, which forces readers to pause and switch exercises, creating unnecessary confusion. If these exercises are meant to be completed in order, there should at least be a note at the beginning stating this explicitly.

glennj · February 10, 2025, 2:01pm

I wonder why we did that in bash?

I did notice this last year but I didn’t do anything about it.

IsaacG · February 10, 2025, 4:09pm

My guess if we figured it’s pretty simple and the deo version has more value?

BNAndras · February 10, 2025, 9:44pm

I suspect Bash can’t implement Resistor Color without some retooling since we’re asking for both the value of a resistor color and a sequence of all the colors. The Bash test runner can only do one or the other at the moment. If we drop the second requirement, it’s basically Resistor Color Duo but with only a single band so I’m not seeing this as a priority except for maybe completeness.

IsaacG · February 10, 2025, 9:51pm

bash now has Resistor Color . Glenn acts fast.