General feedback

[azala]

I met you at the JS meetup last week and you showed me Lambda Explorer.

I'm surprised it got no attention on Hacker News when you posted it 4 years ago. Maybe it sucked back then. This sounds exactly like something they would like and it's better than every other lambda calculus interpreter/tutorial I found in a google search.

I finished it up to the 3 challenge problems (still working on them), here are my notes:

• Problem 3 - comprimise -> compromise
• Dark mode?
• Back/forward arrows too small, hard to find
• The "leftmost outermost" beta-reduction description is confusing as hell and got me tripped up trying to understand the Y-combinator reduction, in particular why (λg.(λx.g(xx))a reduces to λg.g(aa) and not λx.a(xx). A stackoverflow comment that helped me a lot with it was https://stackoverflow.com/a/34305160 because it showed the application order in boldface within each step.
• Problem 25 - I would prefer that the representations of 1 through 4 were hidden. 0 probably should be described in the right half as being identical to FALSE. I covered the right half of the screen for this one.
• Possible to use _ as an alias for "the last thing the interpreter returned" like in python?
• Refreshing saves tutorial progress but not variable declarations

Extremely nitpicky things that you don't have to fix at all

• Shorten FALSE -> F and TRUE -> T, somewhat easier to read and use. I used this shorthand when working with them on scratch paper.
• I was expecting the interpreter to tell me that λa.(a FALSE) TRUE was the Church NOT like it does for Numerals and Booleans. But then again, I guess it's consistent in that it only does it for things we usually think of as primitives.
• λε₁.ε₁ errors out. I know we're only supposed to use a-z and a number, but it is something that the interpreter outputs, so people will probably try to copy paste it as an input while playing around with it.

Again, these are all nitpicks and you did a great job! I don't think I would have bothered learning lambda calculus if the interpreter were not this fun to play with.

[evinism]

Oh hot damn, these are a good list of suggestions!

Let's see if I can respond to all of these: 1: Thanks, i suck at spelling

2: I've never considered! Should be super easy and rather pleasant. Will do, because it takes like 0 time to implement.

3: Will def increase their size. I'm considering doing a small restyle for them anyways, where they take up the entire width of the side panel. Should be a lot more obvious.

4: Noted, I'll try to do a revamp of this. One of the things the tutorial has been severely lacking is diagrams. It would help SO MUCH with the first few problems, and it'd certainly be helpful for normal order explanations too.

5: As in, moving all but the 0th church numeral to be secret, I assume? I'm a little surprised with this piece of feedback, to be honest. I figured showing all the representations in front of you at once lets you see the pattern really clearly.

6: Sure, why not!

7: I've been considering some way of doing this for a while, along with also seeing what's been defined already. I'm thinking the best solution is to just persist everything and give granular options to clear everything in like a gear icon.

As for nitpicks: 1: This makes a lot of sense, and I have no reason to not do that, except MAYBE it being weird with the all-caps w/ the lexer.

2: This is technically undecidable. I could do it for either a truth table or for a few standard forms, but i decided on just not doing it for functions.

3: This is certainly on my radar. It's blocked on a number of things, most recently #60. The point is, my method of avoiding collisions is by using a special character that users can't input, and globally increment a counter. This is obviously a garbage solution, but solving it requires the engine to be aware of what's in scope vs. not. This is by far teh biggest limitation of the engine so far and solving it allows for a lot of neat things.

Thank you so much for all this feedback. This has to be the best group of suggestions I've ever received-- you've obviously put a bunch of thought into this.

[azala]

Re: 5 - Once you have the representations for 0 and 1, the rest are trivial to derive.

As a problem-solving masochist, I wanted to try and play around to sort of "guess" what the representation of 1 was, given the representation of 0 and everything I had learned to that point.

It's probably fine if you leave 0 and 1 shown and the rest hidden.

[evinism]

Dark theme pr here: https://github.com/evinism/lambda-explorer/pull/63, gonna merge momentarily

[azala]

Dark theme works great except links (still blue/purple)

[evinism]

ah, for sure-- not sure quite where you're talking about; might be browser specific. Wanna make a separate issue documenting it? (e.g. exactly where / what browser)

[azala]

https://github.com/evinism/lambda-explorer/issues/69