6:15 map_err should take a closure. Also prefer returning the raw error and mapping it externally with the framework so the actual error could be logged.

@@nathan44u Thank you so much - wild that that example is from the official docs and compiles fine!

in 6:15 , the map_err isn't an error, it's a function so you can pass it instead of creating a closure

@@netfri25 oh great! Good to know the poem exmapl WAS right!

what is the http cli you are using to test the apis

All apps probably need a little db, dive in!

It’s a powerful macro system

@@azaleacolburn Sure is! My video on the macro system is here, if you're interested kzfaq.info/get/bejne/g72Cg7xz1KmpkY0.html

@@NoBoilerplateI know, I've seen it multiple times. Love your content, keep up the good work!

@@azaleacolburn thank you so much! 🙂

Markdown and RSS literally run my life. it's been 10 years already! We need him now more than ever.

@@NoBoilerplate Seconded. My daily notes are in markdown and RSS is how I get my podcasts. His story really reads like a classic tragedy in a modern setting. While trying to not make things too melodramatic: In a bittersweet way, among the many things wrong in our current culture, he is a reminder that there are good people out there. And, even if they get overwhelmed or screwed over, their good deeds can still go a long way.

Who is aaron? I dont want to be disrespectful, just have no clue.

@@fumano2679 boingboing.net/2013/01/12/rip-aaron-swartz.html

His legacy is tainted by what Marak did with Faker.js and Colors.js - distributing malware through open source channels is not an effective form of political protest

YES! This is how I felt after learning Haskell at the start of my python career!

Me too! I've started using generics much more than overloads. I use the rust naming conventions (such as snake_case) for custom vars, and camelCase for stuff that behaves like built-in JS stuff. I've learned to use some FP constructs to avoid using `let` instead of `const` (such as IIFEs, while we wait for `do` expressions). I've learned to modularize my code, and I even write "boilerplate" to make my code safer and more readable (such as implementing a basic `range` generator, to do for-loops like in Rust and Python. Of course, I only do this for "small projects" where importing a whole lib is inconvenient). And more! Learning Rust literally changed my way of reasoning about programs, and it changed my life (not in a *EXTREME RADICAL* way, but considerably enough)

You're too kind!

I think they're not known or consumed as much as they should be; needs more exposure. However, I think they're rated very highly by the community. All the top Rust voices seem to speak very highly of this content, as they should.

@@theherk that's so wonderful to hear!

@@theherk No untruthful words here. The content is great!

Thank you so much! You're very kind to say

Thanks! A few people don't like it, but most do! I like to always know how far through a video I am, and touching/clicking on the screen all the time doesn't do it for me!

Thank you!

SO good right!

Oh I'm familiar, it's not mature enough for me to recommend yet. But you bet I'll talk about it when it is!

Yes! I’d love to hear your thoughts on Leptos. I love it, I’ve been using it for a product that I plan to launch later this year, I think it’s plenty stable!

@@JOHNDATGOONV2 That's the thing with Rust code - it's pretty much all stable! XD Joking aside, I think it's a great tech choice for a project you'd launch a bit later on, perf!

My pleasure, thanks for your nice comment :-)

THE PLAY BUTTON IS ARRIVING NEXT WEEK :-D

@@NoBoilerplate WOOOOO!!

SOUNDS LIKE A NIGHTMARE XD

@@NoBoilerplate honestly from what I remember, it's relatively not bad, and I really liked the highlevel APIs and DSLs, imo (still) they are close to the best thing you can get in Java (ew, Java) However, there was a moment when I committed something to master that broke the tests (yes we rebased-and-committed straight to master) and the guy was almost yelling at me for not running them before pushing (I forgor 💀) and I was trying to suggest "um we can setup like an automated thing that would run the tests whenever you push something [to master]" - not yet knowing at the time what CI is or what are the better practices of it omg - and they didn't have that, and I just looked at the activej repo - they still don't

@@necauqua to be fair, most companies just started investing into CI/CD in the last few years, as it only started to proliferate with stuff like gitlab. I know the more wealthy java houses I worked with decades ago had their own build pipelines already with CVS and SVN, but testing was still often done with manual processes or only around releases.

Wonderful!

Search and ye shall receive crates.io/crates/axum-swagger-ui Of course, they're not first-party, which was my delight when finding poem, but I bet there's others!

@@NoBoilerplate LETS GOO, I will check it out

I’m using utoipa with axum in a project. It is framework agnostic and have similar macros as shown in this video.

@@jakobankarhem6281 that sounds interesting. especially since axum-swagger-ui seems to only adding swagger ui, but utoipa allows decorating the views.

Totally understand, it is a bit intimidating! Whenever you want to learn, here's my video for teaching yourself Rust: kzfaq.info/get/bejne/aM6Igcdn3Z_Xiqc.html

@@NoBoilerplate thank you :)

You're too kind, thank you so much!

No betting needed - there's different frameworks for different purposes! Axum is indeed very good.

it's SO lovely to work with!

You could literally use Poem-openapi for that, as I say in this video, Rust is Cheap: kzfaq.info/get/bejne/asqmmZmcsp2UoYE.html

the first-party openapi plugin - it's lovely!

Nice! There's no warp first-party openapi plugin, but people have told me about docs.rs/utoipa/latest/utoipa/ It's less DRY than poem-openapi, but should be a good starting point!

You are very kind! I have practised public speaking a lot, starting when I was at university - practice makes perfect! I also practised making a little weekly fiction podcast for 3 years before starting this channel. I talk about this all in this video, if you're interested: kzfaq.info/get/bejne/nOB-nsJi3bKte30.html

Thank you!

Same - I'm curious about the differences between Poem and Dropshot, I'll have to try them both at some point I guess

Also, forgot to say that the Poem openapi was much nicer than Utoipa as the schema was defined as code instead of alongside the code 👍

That's the big difference, poem-openapi is extremely DRY

Yeah, the docs could use some work (maybe you or I should do that!) but the examples are great, that's how I learned.

Oh it would be TERRIBLE to do this in any other language. But Rust is so comprehensive that it can do these impossible things, see my video: kzfaq.info/get/bejne/g72Cg7xz1KmpkY0.html

@@NoBoilerplate I understand, but what happen if I don't have a local database? it wont compile? so if I'm working at a large company with a large app/db, and I don't have access to replicate a local database (because there's a highly confidential layer of security and you can only do a query to a sandbox limited db where you need to access via website), I wont be able to code? It may be an edge case, but it's based on the company I'm currently working on, it's a pain in the ass to work with sql queries.

​@@LuisCassihthere's sqlx offline mode where it checks and save the result to a metadata

@@LuisCassih It's not an edge case - CI would suck if you had to have a db in there right? SQLx has this covered, I talk about it in detail here in the video I pinned when I introduced sqlx, here: kzfaq.info/get/bejne/gK-mfsqE3tzYqH0.html

oh, that's a nice feature, I can even create schemas without even accessing the db. Thanks both of you for the insight.

Since this video, the community has rallied around dioxuslabs.com/

Interesting! I wasn't aware of error-stack, probably because it's just a year old. I trust anyhow a LOT, it's the #1 results handler in Rust, and so Eyre being a fork I don't have to think too much about before recommending. I will follow error-stack with great interest, thank you!

@@NoBoilerplate btw, error-stack has full support for Eyre and anyhow! You just need to enable the feature. It's based around the same idea of building up a report from context, except the context is fully typesafe so you can use e.g. enums as errors

@@tabiasgeehuman oh INTERESTING! Well thank you I'll take a look 🙂

Rust library selection is the ultimate ADHD time sink. lol

Yeah, there's a 'winner' but they're ALL wildly fast, that's why I put the normal frameworks down the bottom for context :-D

Come ask those questions on my discord in #newbie-advice, folks will have great suggestions! I have answers: 1. Enums: kzfaq.info/get/bejne/sJNgYMWTyJ6UeoE.html 2. wasm inside Tauri (or electron) kzfaq.info/get/bejne/hpp8gMmcq7K2o30.html

Already did, they do this with the WILDLY powerful macro system, my deep dive is here kzfaq.info/get/bejne/g72Cg7xz1KmpkY0.html I also talked specifically about how SQLx works in this video kzfaq.info/get/bejne/gK-mfsqE3tzYqH0.html

I just can't get too excited for htmx, it's easy to get going but suffers from the same unsafety that js does. Try Yew.rs, it's very simple and will give you huge confidence in your frontend!

@@NoBoilerplate nah it's internal tools only nothing exposed for the web. Well you could completely go the SSR HATEOAS route but htmx reduced the boilerplate/templating a lot.

Oh for sure javascript has a big headstart. HOWEVER we're all getting very excited by leptos.dev - have you tried it?

@@NoBoilerplate I haven't. I'll be sure to check it out when I have time. Thanks for sharing!

the only limit is your imagination! For example crates.io/crates/silicon

Thank you for being chill XD yes indeed! This video isn't about how rust is the best (see my others for that!) but just putting Poem in context.

Why didn't YOU do this comparison before making this comment? I gave you the link in the video! Both Fastify and Hono are extremely mediocre: web-frameworks-benchmark.netlify.app/result?asc=0&f=fastify,hono,express,poem

Fair, though isn't memorisation writing, of a sort? Only writing to the medium of your mind, rather than paper - Dijkstra didn't mean the physical act of writing, really, he meant the act of creation. Consider the sentence "I wrote a song", you might physically write the lyrics, but the tune is in your head.

That'd be a good trick! Here's how I evaluate rust crates, I use lib.rs: lib.rs/crates/poem Note the statistics right at the top of the page: - latest release July 23rd 23 (before then, Feb 23, Dec 22, Nov 22 etc) - #9 in 'http server' - 48k downloads/month - used in 116 crates lib.rs surfaces exactly what I need to know, and all without javascript! Rust and the Rust community make STAGGERINGLY stable software - even 0.0.2 crates I trust more than 2.0.0 npm packages!

That's Fira Code! Though there's a few ligature nerd fonts around

@@NoBoilerplate thanks I’ll have to check it out. I’m using a custom NF font with ligatures but it doesn’t have that ligature

You're welcome Dennis

SO good right! cargo-watch is fine, too, though it flickers when refreshing.

Ooh, you're gonna love the solution: It's got a built in cache mode for CI. Check my explanation here kzfaq.info/get/bejne/gK-mfsqE3tzYqH0.html

Two reasons: 1. compile time verification, 2. optimisation. 1. Only SQLx (I believe) with the query_as! macro can check your queries at compile time. Though it'd be more ergonomic to write in an ORM, I happily give it up to get compile time verification. 2. Every project that uses an ORM eventually has to hand-optimise their queries at scale. Universally, with no exceptions. I am fully bought-in to Rust's "upfront the work" mentality, and writing queries in sql, rather than sometimes writing orm and sometimes writing sql is MUCH cleaner. Also, it's very likely you'll write your own functions to abstract both the ORM or the raw SQL, so you might as well get compile-time guarantees!

no such thing as breaking changes with Cargo, right? Human error, sure XD

1. I don't think we need to 2. Good luck herding the cats XD

​@@NoBoilerplate hahah fair enough

Amazing!

there are 423 crates on crates.io that mention graphql - Go forth!

I've used async-graphql and juniper. I slightly prefer Juniper, although I think async-graphql may have more features/be more active. Both have been more pleasant to work with than their Typescript counterparts.

Leptos looks really good, and I look forward to it maturing!

Poem has built-in i18n support, check the readme! crates.io/crates/poem

@@NoBoilerplate That's nice, but I mostly write desktop applications. I just found a crate named i18n-embed-fl that looks promising. I'll try to understand how it works and if it does what I need.

@@yondaime500 I think you've got it - the standard seems to be crates.io/crates/rust-i18n

Gotcha!

I learned Rust from Amos's amazing posts!

@@NoBoilerplate I'm a big fan too :)

I actually have a whole video explaining my tips on how to learn Rust, and it centres around Rustlings, a code kata system like leetcode problems! Do watch (and check out the other videos in the playlist if you have time kzfaq.info/get/bejne/aM6Igcdn3Z_Xiqc.html

Thank you so much, it's so wonderful to be, effectively, teaching professionally, I've always wanted to!

We're all very excited for leptos! Not quite mature enough for me to be recommending just yet, though.

Every other video I make is on Rust at the moment :-)

We need him now more than ever

Looks like miette is on the `thiserror` side - creating good library errors, not handling good library errors. I don't write libraries, but if I did, I'd check it out!

​@@NoBoilerplatemiette is on all sides: it has guides for using it in either libraries, applications or as the main() return. it's main claim to fame is ultimately very nicely formatted errors, akin to rust compiler errors, which are desirable in applications too.

@@laundmo it's very exciting! Thanks both for bringing it to my attention!

An excellent point. Yes, especially for API versioning, you'd want that decoupled, which is easy enough, just make another struct. The strict typing of your API, and the compiler keeping you safe when making changes, still applies after this useful weakening of the relationship between DB and API.

It's the stand-out best SQL database on the planet, and has been for my entire 15 year career.

crablang until it's merged, maybe. I remember the node/io.js split - history is cyclical XD

Working with an ORM - like Diesel in the Rust ecosystem - might be a good fit if writing SQL seems too daunting at first.

Diesel is a very good ORM, but it doesn't do compile time verification, and its runtime errors are not very nice.

I too used to feel like this. You could try Diesel, a very popular rust ORM. Here's why I now prefer to write SQL: In every large project I've worked with, we've EVENTUALLY had to write SQL. ORMs don't scale, and the bigger your app gets, the more hand-crafted SQL you have to write. This is the same problem as Javascript, it STARTS OFF nice and simple, but by the end, you wish it weren't so simple, because the edge cases break you. Use SQL with SQLx, and don't be afraid, because the compiler is holding your hand :-)

Planetscale has a good intro to SQL. They also have a free tier for hobbyists.

@@NoBoilerplate that makes sense, but an ORM definitely sounds like a much nicer starting point. Maybe I'll also look and see if there's a kata-like course for SQL! I used Tom Kunc's MacroKata to learn Macros in Rust and it was awesome

Ah, so do I. Nim was my gateway from Python to Rust, igniting my passion for single-binary deployment and static typing! Popularity matters, as I always say, so I looked at three mainstream languages that I stood a chance at writing professionally: Go, Haskell, and Rust. Go is ruthlessly practical, but with no beauty, Haskell is beautiful but not practical, and Rust has the best of both, and is somehow greater than the sum of its parts! I wrote this comment on a hackernews nim v rust comment thread last month, perhaps it's useful to read: Nim is too timid. Most languages are too timid. My context: I'm a maintainer of iNim, creator of the Rust channel, No Boilerplate, and a professional python web developer for 15 years. The features you are excited about in Nim are compromised for the sake of simplicity: - Macros can't change syntax, - There's a limit to how real-time the standard language is, - Compiling to javascript is a nice trick, but if it didn't rely on GC it could compile to webassembly, which is better, - Ask any advanced python developer: Indentation for syntax is a bad idea at scale. Hell, have you ever used YAML? What Rust does is a revolution because they solved the problem of memory safety without garbage collection with the Borrow Checker. The knock-on effects are GARGANTUAN. This was a hard problem to solve, and it makes the language more complex to learn, but it unlocks all the language's superpowers. If you 'cheat' with a GC, you get nothing. Nim started its life, just as most languages did with a GC, and the efforts to keep the syntax the same while offering alternatives are greatly compromised. It's far easier to add GC/RC to a low-level language than it is to remove it from a high-level one, where the whole ecosystem relies upon this behvaviour. Of COURSE I know about Nim's plugabble GC options. That doesn't impress me at all (though Araq's effort is to be commended). In Nim they are compiler options, in Rust they're libraries. The difference is vital. For example, at a per-variable level, I can choose to reference count with doc.rust-lang.org/std/sync/struct.Arc.html if I so wish (I rarely need to because the borrow checker is magic). If you want python, you can have it inside a Rust macro, sort of like this crates.io/crates/inline-python because rust macros CAN change syntax. With Macros and no GC, you can rebuild ANY FEATURE inside Rust, the floor of the abstraction is the metal, and the ceiling is the sky. This is what Rust is all about: full power to the developer, no compromises. My video on this topic for those interested is: kzfaq.info/get/bejne/htt9i7qUld7We2Q.html I love both languages, but I can only see one language being the standard next 40 years, and demonstrably so can Linux, Microsoft, Cloudflare, Discord, and many others.

@@NoBoilerplate Thank you for the response, and thank you for making it a wonderful response! I feel like I've met a celebrity! I've been loving your videos and sharing them wherever I can. ☺️ The way you reference Nim just caught me off-guard because I had just discovered it and was really liking it. I'm primarily a Python developer, so it was really nice to see similar syntax and I appreciate this simplicity over the very verbose Rust I've been trying to learn. And have you seen the Uniform Function Call Syntax? ♥️♥️♥️

@@LoganVanCuren UFC is SO fun isn't it! Look, learn nim, do it without any guilt, it's a great language. Rust will be here when you want more ;-) BTW Rust has something that feels a bit UFC - you can attach methods to any struct using `impl`. It feels like much looser coupling than classes (though less loose than UFC) doc.rust-lang.org/beta/std/keyword.impl.html

We need Aaron more than ever

We need him now more than ever

I really like Nix, and have NixOS on a laptop for testing. However... This whole stack is pure Rust, even using RusTLS, no external dependencies other than rustup.rs. Nix is ESSENTIAL in other languages, but I've found since switching to Rust, I can just choose pure Rust dependencies and never trip up. As Cargo is extremely deterministic, I don't miss Nix there either. The Rust Editions system guarantees anything that compiles today will compile in 40 years time!

I'm uninterested in closed source silos.

Axum is great! They're all great really. As I said in the intro to poem, axum doesn't have a first-party openapi plugin, and I've heard that those it does have are not ideal.

@@NoBoilerplate The reason I like axum is its well integrated in tokio stack. So I think it will be maintained for a long time. Also having OpenAPI is not important to me, that's just my prespective, its a developer feature mostly and they usually use Postman or something similar. Again great video and I hope you find success now that you are doing rust videos full time. I would love a detailed video on tracing, not a gloss I think basics everyone can get.

@@AbhishekBajpaiHere Thank you so much! Axum's great too, of course. You can't go wrong with the Rust frameworks! What's to know about tracing? The readme's very comprehensive - what value could I add?

Solid name honestly. I learned Rust from Amos at fasterthanli.me :-)

We need his help now more than ever.

We need him now more than ever

Apologies! This isn't targeted at beginners to Rust, however I do have a whole playlist that IS! Here's the exact video in that playlist where I talk about bacon kzfaq.info/get/bejne/mqyberGWmdXFe58.html

@@NoBoilerplate Ooh, thank you! Love your podcast too btw!

@@tiagomacedo7068 oh thank you so much! I started planning Lost Terminal season 14 just today! Drops on the first Monday of November 😀

Thank you so much!

Diesel, like all ORMs I've tried so far, does not connect to your real sql database at compile time and validate your queries. Just incredible, check out kzfaq.info/get/bejne/g72Cg7xz1KmpkY0.html

@@NoBoilerplate but you can provide file with sql commands so you will have structures of tables, so you will have some compile time checks like for types

@@danyldavidiuk9507 That is not good enough

Watch "The Internet's Own Boy", linked in the pinned ERRATA comment.

Yeah, if you check the link there are some nim and cpp webservers that punch a bit higher than ActivJ, but as I said, you'll never persuade a team to use them, so I feel safe discounting them XD I'm here talking about Rust not just because it's a WILDLY great language, with alien magic in, but that, somehow, despite all this, it's already the #19th most popular language redmonk.com/sogrady/2023/05/16/language-rankings-1-23/

There are other metrics than popularity (but rust for http stuff is good)

@@user-du9ch3tn2v popularity doesn't matter comparing the top languages, they're all popular enough, but if you see where nim (for instance) is on that graph, it really does matter.

you MUST have used tokio! really?

​@@NoBoilerplate No, I really never used it. I've seen it a long time ago, but I always thought, I don't want to know that much. I was rather expecting critique on not using Serde :)

If you're talking about the development experience, SQLx has you covered, I explain their 'offline' feature in this video: kzfaq.info/get/bejne/gK-mfsqE3tzYqH0.html

though mysql has a large legacy install base, postgres has beaten it in features and performance my entire career.

Click through, it's comically slow (10x) compared to the big players. web-frameworks-benchmark.netlify.app/result?asc=0&f=poem,fastapi I don't hate python, but fast it aint. I'm DEVASTATED they're keeping the GIL for python 4.0. That's a CHOICE.

Has there been any official talk of python 4.0?

@@murtaza6464 I can't find where I read about it now - but reading the mailing lists they are focusing on optimising so much that the GIL doesn't hold them back. Sure, great, sounds wonderful. Byeeeeeee!

It's because it's actually a frontend for C++, it bypasses all of nodejs and replaces them with native wrappers. SOME might call this cheating ;-) Sifrr is based on github.com/uNetworking/uWebSockets.js You find this a lot at the top of the benchmarks - artificial frameworks that are designed to 'win' framework competitions, but don't actual use your language in a normal way, so you'd never choose them (as you have to throw away much)

I'm as surprised as you

Not quite mature enough yet, but it's very exciting!

We need him now more than ever

Rust is indeed an excellent language, and though you're right that the web frameworks are still being developed, I would trust a 0.2.0 Rust framework over a 5.0.0 JS framework in production every day of the week. Note I say 'trust' not 'use', there are many cases where the downsides of a JS framework (say, react or express) are overwhelmed by the benefits you get from popularity and longevity: If you want gmaps in react you can choose between several plugins. If you want it in rust wasm (leptos or yew, say), you'll have to write that plugin yourself. Some people will be fine with that, others will have to wait, and use React or similar in the meantime. But this doesn't mean that you should write off the whole language, as your comment suggests. That's too blunt. The web stack I feature in this video: poem-openapi coupled with SQLx, TODAY, is better than any other option in any language for building a lightweight API in production. This combination of rust/poem/sqlx gives you features you simply can't have in any other popular language or framework. compared to rails-api or express or django-restful etc etc, the stack I show in this video is more reliable, safer (both memory and high-level) and faster than them all. As a senior web dev, there's one thing above all else that I care about these days: that you can actually finish your project with this stack. It will sit in production, reliably, being upgraded as the compiler improves without a single line of code that needs porting, for decades to come. My team and I can move on to other projects. There's enormous momentum in the Rust ecosystem, I'm testing out leptos for replacing js in my frontend projects, and I'm extremely excited by what I'm finding, for instance. But unlike poem, I don't recommend laptos for everyone just yet, perhaps in a year. What I do recommend is that for people who are sold on Rust, this is a good stack for building an API, which is the vast majority of backend web projects. For now, I recommend to most people to stick a react app or a swift UI on the frontend. But not for long! :-D

Wait till you discover my other videos XD (specifically this one kzfaq.info/get/bejne/aM6Igcdn3Z_Xiqc.html)

@@NoBoilerplate Heh I was being facetious. Watched all your vids, and love them. Does feel though that you love the aspects of Rust that are actually stronger in Haskell than Rust (imo). I wonder if a more FP, and less systems would be ideal

@@dombou70 I adore haskell, learned it 15 years ago and it made me a better programmer in OTHER languages! However, Popularity Matters. I evaluated go, haskell, and rust in 2020 and found go was very practical, but not beautiful, haskell beautiful but not practical. Rust is a wonderful blend of haskell and lisp, all in syntax that doesn't scare people away!

@@NoBoilerplate That's fair. Fwiw, I went through my Haskell phase, and i've also moved on as the community was just too small. I now use TS and it's mostly good enough. I'd love proper ADT's like Rust has, and a std lib built around Result/Option types

@@dombou70 TS is great if you have no option, yeah, I dig it for sneaking in some safety into frontend teams! Yes, this is the exact problem - you can TELL when a language wasn't built around sum types, like TS, if it gets bolted on later, you have deep compromises that you can't remove from the language. RUST IS SO EXCITING FOR THIS

Elaborate hand-crafted native extensions. They're basically built to win benchmarking competitions.