Convert mentor notes to approaches

[colinleach]

See forum for context.

There are a bunch of mentoring notes on the website-copy repo, which were used in earlier versions of Exercism. Now (late 2025), many people have forgotten that these exist, which is a waste.

It would be better to have these as Approaches, so that mentors can easily access them, and they are available to students after completing the exercise.

[depial]

I seem to remember coming across approaches before, maybe on existing exercises? Or maybe I'm confusing that with Dig Deeper content... In any case, they're quite interesting, so I think it'd be a good idea to take a look at adding them.

[colinleach]

Approaches are one of the things that show up under Dig Deeper.

[colinleach]

There may be less needed than I first thought. I checked what was on website-copy (left column), and also did a search for existing approaches in exercism/julia (right column). Not a perfect match, but quite a lot of overlap.

Mentor notes	Approaches
atbash-cipher	atbash-cipher
bob	bob
clock	clock
custom-set
difference-of-squares	difference-of-squares
	etl
hamming	hamming
	largest-series-product
leap	leap
luhn	luhn
nucleotide-count	nucleotide-count
pangram	pangram
phone-number	phone-number
pythagorean-triplet	pythagorean-triplet
raindrops
	resistor-color-trio
robot-name	robot-name
roman-numerals	roman-numerals
rotational-cipher	rotational-cipher
run-length-encoding
secret-handshake	secret-handshake
	square-root
	sum-of-multiples
triangle
word-count	word-count

[colinleach]

I checked back to remind myself what I wrote for Python approaches:

all-your-base
circular-buffer
dnd-character
leap
matching-brackets
pythagorean-triplet
roman-numerals
sieve

There's not much overlap with what already exists for Julia. Obviously, the code would be different, but the algorithms and some of the benchmarking discussion would be relevant. See pythagorean-triplet for an example of how to waste a week going completely OTT with approaches (not recommended!).

[depial]

See exercism/python#3582 for an example of how to waste a week

That is an admirable amount of effort!

It's nice to see that there's so much overlap. I've only had a peek, but what do you think about the content of the current approaches? They seem to be made up of an introduction.md and a config.json. Is this the format we'd be looking to emulate?

[colinleach]

They seem to be made up of an introduction.md and a config.json

That's the minimum format, with both files required. Optionally, we can add per-approach subdirectories, to discuss an individual approach in more detail.

For serious benchmarking, the introduction.md only contains summary info. Code, results and more detailed discussion would be under .articles/, not .approaches/.

Coincidentally, I was just starting to look at the existing approaches as you posted this. More to do on that, but they seem pretty short and clear.

Less positively, I spent some time this morning playing with the raindrops mentoring notes and seeing if I could easily convert them to Dig Deeper approaches. Short answer: no, the current document is highly technical and a long way from being suitable for something student-facing in a difficulty-1 introductory exercise.

[colinleach]

what do you think about the content of the current approaches?

I haven't looked at all of them, but I've been pretty impressed with what I've seen so far. Of course, anything can be improved if necessary, but I'm inclined to tick these off as something we don't need to worry about.

The 4 exercises with mentoring notes but no approaches document are quite a mixture.

• custom-set: Not much more than an empty template, so of little use to us.
• raindrops: Too scary for students in its present form. We could probably steal some content to use in a major rewrite.
• run-length-encoding: A good starting point, if we break it up into sections and add more explanation.
• triangle: Needs reformatting, probably with more explanatory text, but a good starting point. There has been a lot of recent (interminable!) messing with the canonical tests, so we need to check if it's up to date.

I checked back to remind myself what I wrote for Python approaches

I'm tempted to try porting matching-brackets to Julia. I've no experience with Julia benchmarking (and its many traps for the unwary), so I might learn something. Also, the Python results surprised me, and I'm curious to see if the approaches have the same order of performance in Julia.

This is a more complete list of possibles:

• all-your-base
• circular-buffer
• matching-brackets
• sieve

A bit of mathematical bias there, one might think.

[depial]

I might be able to help with content for custom-set since I had a pretty thorough mentoring session for that one where I learned a lot.

In general, I wouldn't mind helping with these in any number of ways, because I like to try to take different approaches when I do exercises anyway. Here it might not be such a big issue if there's authoring style differences, so let me know where/if I can be of help.

I've no experience with Julia benchmarking

I have fairly limited experience, surely not much more than nothing :D

[colinleach]

I just submitted raindrops to get us started: PR #1050.

If I do more of these, I'll concentrate on the 4 exercises I already did for Python. I'd be delighted if you take some or all of the mentoring notes conversions.

[depial]

I'll try to take a look and get a feel for them. I might be a bit busy for the next few weeks, but I usually find ample time to procrastinate, so I'll try to channel that here :D