Copying the Ruby track structure

[iHiD]

Hello @exercism/python

Python is one of two tracks that we're really struggling to mentor. The main reason seems to be that the structure of the track isn't massively easy for beginners to upramp on, so mentors are having to solve a lot of issues on each exercise.

We're currently working on a project to build a system to make structuring tracks easier. We'd like to work on the Python track for that as soon as it's ready. However, in the meantime, I think we should re-order exercises to be the same as in Ruby, which has had a huge amount of effort put into it in terms of thinking about structure.

If anyone massively disagrees then please speak up. Otherwise we'll do this in the next week or so. If anyone would like to volunteer to do the PR for this, then that would be super helpful, otherwise I'm tentatively assigning this to @kytrinyx .

[yawpitch]

A lot of the exercises in the core Ruby track will have a significant amount over overlap in the Python track.

Personally I would eliminate two-fer entirely; I'd rather see the ability to handle a default argument be a bonus challenge on hello-world, and plenty of the other exercises have string formatting and concatenation.

I would ten to put isogram before acronym,

The raindrops exercise has already been deprecated in Python and I wouldn't bring it back. It again has a lot of overlap with exercises above it and we really don't want to front-load just string methods in all the exercises.

I would suggest we keep leap and book-store and binary-search, while I wouldn't consider twelve-days or any of the other "song" ones for inclusion.

[iHiD]

@yawpitch Superhelpful. Thanks. The point of two-fer in Ruby is to teach the very simple things like indenting (two spaces), conditionals, etc. The default argument is in many ways incidental to the point of the exercise. Is there another exercise that is as simple as two-fer that could achieve the same thing?

[yawpitch]

Also I'll add in a vote for sieve and run-length-encoding both being better core exercises than rest-api, which I would argue really isn't ready for prime time, much less core.

[yawpitch]

@iHiD though I personally dislike bob, I think it covers all those bases for Python; it requires proper indentation and introduces simple control flow with if and elif; I'd say it's certainly more useful than two-fer at weeding out those issues.

[yawpitch]

I'd be happy to see kindergarten-garden and allergies leave core ... they both soak up quite a bit of time and again have a lot of overlap one with the other ... if I were to keep either it would definitely be allergies because it introduces bitwise operators -- quite possibly the least reached for tool in the Python standard lib -- but I don't think that's enough to save it.

[rpalo]

Personally, I'm OK with Two-Fer -- it's a good confidence builder, but it's weird that it's so far back in the course. Should probably be right after Hello World.

While I like both Kindergarten Garden and Allergies, I agree that they're very similar. I agree that allergies would be a good one to be core, because if it was a side-exercise, I'm pretty sure a big segment of the students who try it would miss the binary entirely. Kindergarten garden's gotchas could be picked up as community comments on a side exercise probably.

I personally really hate mentoring Book Store. However, it's a good exercise, I enjoyed it when I went through it on the track, and I really benefited from having a mentor there. But maybe it would be ok to leave as a side exercise that people could opt into?

I really like Markdown and REST API as core exercises. They're kind of like the track master's theses, or, at least, that's what it feels like as both a student and a mentor on them. They present a little bit of design/architecture thinking as well as the idiomatic Python stuff.

I'd have to comb through the side exercises to see if I think they would be good core ones, but, overall, I think you probably couldn't go wrong with mirroring the Ruby track for most things. It feels well-laid-out and each exercise introduces something important and new.

[cmccandless]

Just in case there was any lingering concerns about this from a maintainer's point of view:

When I reordered the exercises in this track for the new Exercism, I based the ordering mostly on the C# track. I think Ruby is a much better fit for a starting model than C#, so I have no problem with a significant reorder.

A few other thoughts:

• hello-world should remain as simple as possible (no optional parameters)
• two-fer has always seemed a little redundant to me
• rest-api still needs a good deal of work, and should probably not be a core exercise (for now)

[iHiD]

Personally, I'm OK with Two-Fer -- it's a good confidence builder

This is my strong feeling too. I feel like it is a nice gentle step up from Hello World, gives mentors the first opportunity to iron out newbie problems (inconsistent indenting, weird variable names, etc), and in 50% of cases can be approved in 20secs.

[cmccandless]

in 50% of cases can be approved in 20secs.

Do we actually have/track this metric (or is it anecdotal)? If so, that's awesome!

[iHiD]

It's anecdotical but I've mentored >50 twofers and so I think it's a pretty authoritative anecdote :) @kytrinyx has similar experience in her tracks and @ErikSchierboom in his.

[yawpitch]

I suppose my issue with two-fer is that it's basically hello-world with the added twist of a default argument ... if we actually verified that hello-world passed the test we'd eliminate most of the problems you're describing as well, and such problems are currently being addressed in leap with just one iteration as is.

That said part of my issue with it is that -- where it's currently placed in the order -- if it takes more than 2 seconds to approve, it's too much.

[iHiD]

Yeah. When we moved it to after hello-world in Ruby, it fixed so much stuff, because it made the first interaction for students and mentors very pleasant and easy. I actively look for two-fers at the end of a hard day, because I know that I can either be positive and encouraging, or immediately helpful to someone who is starting out from a pretty beginner position - both of which feel nice.

Leap is good too for its simplicity - it's just much harder as a learner as there's actually something to solve that's logic-based, not code-based, and it hurts my brain every time to look at as a mentor still :)

[yawpitch]

Right ... personal bias then. I'm now pretty certain I'm among the world's foremost experts on "what is a bloody leap year", but I'll readily admit I'd prefer to not have to walk so many complete beginners through 6-7 iterations on it, with 3-4 just on basic setup issues.

I remove my objection to two-fer ...

[cmccandless]

I remove my objection to two-fer ...

Seconded. I don't currently mentor, so I don't have that personal perspective. Thanks @iHiD.