`pretty_diff` - ⚖ pretty printing the difference between Gleam values

[quartz estuary]

I've published a package that can be used to pretty print a rich diff between two Gleam values
For example:

let jak_likes = ["FP", "Gleam", "programming"]
let jak_langs = dict.from_list([#("Italian", Native), #("English", C2), #("Spanish", A1)])
let jak = User("Giacomo", 25, jak_likes, jak_langs)

let tom_likes = ["OOP", "Java", "programming", "playing the piano"]
let tom_langs = dict.from_list([#("Italian", Native), #("English", C1), #("Japanese", C1)])
let other = User("Tommaso", 25, tom_likes, tom_langs)

pretty_diff.from(one, and: other)
``` 
Will result in this pretty printed diff!

https://github.com/giacomocavalieri/pretty_diff/tree/v1.0.0

@umbral quail would you accept a PR to use this in startest? 🥺

[pulsar flax]

Awesome!!

[umbral quail]

I would like to try it out!

Although I think I'll want to play with it myself to make sure I'm happy with the output

[quartz estuary]

Lmk if there's anything you wish it did differently

[pulsar flax]

It looks a lot like the clojure deep-diff2 output

[quartz estuary]

It was very much inspired by that!

[pulsar flax]

i figured ...i will have to try this later today for the quickcheck library

though it would be nice if it could be included in gleeunit somehow

[umbral quail]

I don't think Gleeunit is getting many updates these days 😅

[pulsar flax]

lpil has said they are not accepting new stuff for it on other github issues and stuff

[quartz estuary]

@whole salmon would you accept a PR to add this to gleeunit?

[pulsar flax]

im interested in startest at least

especially if you put the diffing in

[lament wadi]

i'm curious how this compares to gap: https://hexdocs.pm/gap/

pretty_diff seems more generic in that it does all the things?

[quartz estuary]

Gap only works with lists and strings, pretty_diff works with any gleam value

[lament wadi]

okay cool that's what i figured

i dig it

[quartz estuary]

gap is way more useful in general because you have access to the Diff data structure

While I intentionally only expose the from function that returns a String and not the internal representation of diffs

[lament wadi]

interesting tradeoff

[broken silo]

are you using this as part of birdie?

[quartz estuary]

Yeah I want this to be very limited and opinionated to avoid people doing crimes with the internal representation of diffs 😁

[lament wadi]

perfect for me who would just want something simple and straightforward

[quartz estuary]

Nope in Birdie I'm using a textual line by line diff because I'm comparing text blocks and not gleam values directly

[lament wadi]

because yes if you make crimes possible, i will commit them

[broken silo]

but could birdie also take a gleam value instead of a string and then use pretty_diff? different function simply?

if that makes sense at all

[quartz estuary]

Well, that isn't possible because the comparison is made with the textual content of the snapshot, not another in-memory value

[broken silo]

oh that makes total sense!

[quartz estuary]

And even if that were possible I wouldn't want birdie to ever take something that is not a string as input, the less magic hidden inside birdie the better!

I think there's a lot of advantages in birdie only allowing the content of a snapshot to be a plain String

[broken silo]

it is fine, I love pprint+birdie

[lament wadi]

i agree, though i'm sure someone is gonna come along and just string.inspect piped to birdie

[broken silo]

btw would you consider accepting a PR that auto sets the snapshot name to the test function name it is run from (prefixed by the module name/path)?

[quartz estuary]

Oh yeah we have pprint for that, but that's fine because you are in control and you have to explicitly opt-in that magic behaviour

[broken silo]

I mean for the testing it is good enough - for refactoring it is less great imho

[quartz estuary]

Nope, I'm still not sure this is the best approach and it would be a breaking change

[lament wadi]

https://tenor.com/mdLgqtIIp0i.gif

[broken silo]

in case you want to do it, let me know, it would just be a different function name, so not breaking?

[quartz estuary]

Ah so you mean another snap function

[broken silo]

yes (as we do not have optional args :D)

[lament wadi]

add crackle and pop so you can reunite the trio

[broken silo]

@quartz estuary IN CASE you come to the conclusion that it is worth it (in addition and backwards compatible), let me know and I can try to do the plumbing/chore.

[broken silo]

now after looking at this how it mesmerizes me I finally noticed it!

[whole salmon]

@quartz estuary if we add this to gleeunit and publish it we'll break the tests for glam, classify, and pretty_diff

Dep loop