Series of Gleam Articles

[dusty-phillips]

I stumbled across Gleam a while ago and have found it very entertaining to write in. I noticed there isn't a ton of third-party content on it, so I started writing some. I don't have much Erlang or functional programming background, so I wouldn't take any of this as good practice, but hopefully some might find it interesting, and others might be able to provide some professional feedback:

• Part 1: Hello World
• Part 2: Looping

[michallepicki]

Looking forward to next articles! As I use functional programming day to day I find your insights on Gleam very interesting. For example, I rarely use the iterator module - I usually reach for the map, fold, fold_until and similar functions from the list module, or write recursion by hand. But I think it makes sense that the iterator module interface seems "closer" to for/while loops.

About the Erlang list representation and copying: data can be shared between lists when you're prepending to a list. If your list can be long, prefer prepend ([new_elem1, new_elem2, ..old_list] or list.prepend) instead of list.append (there's a note about performance in documentation there). A common pattern is to build up a list with prepending and list.reverse at the end.

[lpil]

I typically only use iterator if reading the entire input would use too much memory, and instead I wish to stream the input to the output and perform transformations as I go. For example processing lines in a large file.

[dusty-phillips]

I guess I defaulted to looking for iterator because that's what Python encourages. I'll try to do another article on the structure of lists later; I imagine it'll require some research.

[inoas]

Read through Part 1

I believe not having runtime exceptions is very high on the core dev/main devs list/s - so high in fact that 1/0 runs (because its types check out it has to :D) - so I am pretty certain full exhaustiveness checking will happen (other than say C target which onl hopefully will <3) - in short: I’d just write that it is not yet implemented but help is welcome 😉 - otherwise it looks like gleam does neglect that on by design imho (much like nulls in dart 1.0 :/)

I love the way piping / |> is explained to be read as becomes the first argument of.

iodata is now called string_builder AFAIK.

[lpil]

Yup, we will have full exhaustiveness checking #34

[dusty-phillips]

The iodata code was copied from the documentation. I put gleam-lang/website#170 up to correct the docs if that is correct.

[lpil]

Wonderful stuff! Have you shared these on twitter, reddit, etc? I'd love to re-share them, I think people would enjoy them :)

Some notes on the first one:

If you’re not using a Mac, you may have to go to a bit more effort, installing Erlang as a separate step and possibly building the Gleam binary from scratch. The current Gleam compiler is written in Rust, so you’ll have to install an entire separate programming language toolchain.

Rust isn't required to use Gleam, you can generally download a precompiled version from GitHub (https://github.com/gleam-lang/gleam/releases), and we're on a few of the various package managers.

Note that in order to run this program you’ll need to use -- to tell gleam that your remaining arguments should go to your program, rather than to the gleam binary. So, for example, run gleam run -- Dusty instead of gleam run Dusty.

The -- isn't required unless you are passing --flags to the Gleam program, and that is due to a bug in clap. Once they have fixed that -- won't be required unless you're giving a flag that gleam run itself accepts.

In Python, I’d write " ".join(names), but Gleam doesn’t have that (admittedly bizarre) construct.

Check out string.join https://hexdocs.pm/gleam_stdlib/gleam/string.html#join

[dusty-phillips]

Thank you! I'll make a few updates. I actually don't use twitter or reddit, so I've nothing to reshare, but feel free to link to it anywhere.!

[dusty-phillips]

I've created a couple more articles as I become more comfortable in the language:

• Part 3: More on tail recursion
• Part 4: Introducing OTP

[michallepicki]

I noticed links to previous articles lead to 404s, they might need small tweaks

[dusty-phillips]

Should be fixed now. The links Hugo generates when testing locally are different from those when it builds for production. Very strange.

[gofenix]

awesome

[TanklesXL]

Hey great articles! Just a note on this bit:

Let’s now import gleam/crypto and calculate the hash of that password: crypto.hash(crypto.Sha256, password) |> io.debug
This one line example demonstrates why we need to be especially careful of how we design APIs in Gleam. In my opinion, the hash function should accept the password bitstring as the first argument. Then I could have piped the password in as part of the original pipeline.

What you want in terms of data-first piping in this case is actually perfectly doable by using positional piping with _ as follows:

  password
  |> crypto.hash(crypto.Sha256,_) 
  |> io.debug

Although I do agree for something like this the bit string should be the first argument, but it's nice that with _ it doesn't make that much difference for an end-user of the library.

[dusty-phillips]

Oh wow, I can’t wait to try that, thanks!

[dusty-phillips]

Wrote an entire article inspired by @TanklesXL tip: https://dusty.phillips.codes/2022/07/24/a-tutorial-introduction-to-gleam-part-5/ Thank you!

[TanklesXL]

Oh wow, nicely done! 🎉 Glad to have been a source of inspiration