Everywhere: Abandon Swift adoption
After making no progress on this for a very long time, let's acknowledge
it's not going anywhere and remove it from the codebase.
And this was me trying to use Swift for a data access layer + backend web API. There's barely any guidance or existing knowledge on using Swift for backend APIs, let alone a web browser of all projects.
There's no precedent or existing implementation you can look at for reference; known best practices in Swift are geared almost entirely towards using it with Apple platform APIs, so tons of knowledge about using the language itself simply cannot be applied outside the domain of building client-running apps for Apple hardware.
To use swift outside its usual domain is to become a pioneer, and try something truly untested. It was always a longshot.
But there's a lot of hokey, amateurish stuff in there... with more added all the time. Let's start with the arbitrary "structs are passed by value, classes by reference." And along with that: "Prefer structs over classes."
But then: "Have one source of truth." Um... you can't do that when every data structure is COPIED on every function call. So now what? I spent so much time dicking around trying to conform to Swift's contradictory "best practices" that developing became a joyless trudge with glacial progress. I finally realized that a lot of the sources I was reading didn't know WTF they were talking about and shitcanned their edicts.
A lot of the crap in Swift and SwiftUI remind me of object orientation, and how experienced programmers arrived at a distilled version of it that kept the useful parts and rejected dumb or utterly impractical ideas that were preached in the early days.
You can do classic OOP, FP, Protocol-Oriented Programming, etc., or mix them all (like I do).
A lot of purists get salty that it doesn’t force implementation of their choice, but I’m actually fine with it. I tend to have a “chimeric” approach, so it suits me.
Been using it since 2014 (the day it was announced). I enjoy it.
There are plenty of valid reasons to use classes in Swift. For example if you want to have shared state you will need to use a class so that each client has the same reference instead of a copy.
This is the same way that C# works and C and C++ why is this a surprise?
Swift structs use copy on write, so they aren’t actually copied on every function call.
It’s a common misconception that comes from the standard library data structures, which almost all do implement CoW
You can also look at the source code for the language if any it’s confusing. It’s very readable.
Computer languages are the opposite of natural languages - they are for formalising and limiting thought, the exact opposite of literature. These two things are not comparable.
If natural language was so good for programs, we’d be using it - many many people have tried from literate programming onward.
I don't see a case for "complex" vs "simple" in the comparison with natural languages.
Here's a very simple, lexical declaration made more human friendly by use of the preposition `my` (or `our` if it is packaged scoped)...
my $x = 42;BTW, one side benefit of LP is freedom from arbitrary structure of DSLs. A standard practice in LP is to declare and define objects in the spot in which they are being used; LP tools will parse them out and distribute to the syntactically correct places.
It's exactly the opposite. Writing and reading are asymmetrical, and that's why it's important to write code that is as simple as possible.
It's easy to introduce a lot of complexity and clever hacks, because as the author you understand it. But good code is readable for people, and that's why very expressive languages like perl are abhorred.
I 100% agree with your statement. My case is that a simple language does not necessarily result in simpler and more readable code. You need a language that fits the problem domain and that does not require a lot of boilerplate to handle more complex structures. If you are shoehorning a problem into an overly simplistic language, then you are fighting your tool. OO for OO. FP for FP. and so on.
I fear that the current fashion to very simple languages is a result of confusing these aspects and by way of enforcing certain corporate behaviours on coders. Perhaps that has its place eg Go in Google - but the presumption that one size fits all is quite a big limitation for many areas.
The corollary of this is that richness places an burden of responsibility on the coder not to write code golf. By tbh you can write bad code in any language if you put your mind to it.
Perhaps many find richness and expressivity abhorrent - but to those of us who like Larry's thinking it is a really nice, addictive feeling when the compiler gets out of the way. Don't knock it until you give it a fair try!
Problem is, that makes every small part of the program simple, but it increases the number of parts (and/or their interaction). And ultimately, if you need to understand the whole thing it's suddenly much harder.
Surely you can write the same behaviour in "clever" (when did that become a negative attribute?) or "good" way in assembly. You are correct. But that's a different matter.
> Get into a rut early: Do the same process the same way. Accumulate idioms. Standardize. The only difference(!) between Shakespeare and you was the size of his idiom list - not the size of his vocabulary.
Here's a Python rut:
n = 20 # how many numbers to generate
a, b = 0, 1
for _ in range(n):
print(a, end=" ")
a, b = b, a + b
print()
(0,1,*+*...*)[^20]
I wanted to try using a native language other than C++ and Swift ostensibly seemed easier to pick up. I continue to use Swift for iOS app development though where it is much easier to use; but that has its own share of compromises and trade-offs - but not centred around Swift, around SwiftUI vs UIKit.
Personally I wouldn't mind either but my point is that they probably want to cater to the average person, and not just security conscious tech savvy people, and if that's the case, then you really can't exclude FB/IG/YT and others from working properly in your browser.
This kills the internet.
For the record, I don't have a dog in this fight. As long as it runs on Linux, I'm willing to test drive it when it's ready.
It's a shame, I think swift is an underappreciated language, however I understand their reasoning. I think if they tried to just use swift from the beginning it would have been too ambitious, and trying to add swift to a fragile, massive project was probably too complex.
Why not D?
Fascinating.
They've shown the idea it is better on C++ interop is wrong.
I don't know enough to say Rust has same OO support as Swift, but I'm pretty sure it does. (my guess as a former Swift dev: "protocol oriented programming" was a buzzy thing that would have sounded novel, but amounted to "use traits" in rust parlance)
EDIT: Happy to hear a reply re: why downvotes, -3 is a little wild, given current replies don't raise any issues.
Probably the same reason why Rust is problematic in game development. The borrow checker and idiomatic Rust do not go well together with things that demand cyclic dependencies/references. Obviously there are ways around it but they're not very ergonomic/productive.
Said differently: the C++ interop did not support calling the C++ library I wanted to use, so I wrote a C wrapper.
He even made an attempt at creating his own language, Jakt, under SerenityOS, but perhaps felt that C++ (earlier with, now without Swift) were the pragmatic choice for Ladybird.
Rust was born at Mozilla, sort of. It was created by a Mozilla employee. The first "real" project to put it into action was Servo of which parts were adopted into Firefox. While Rust may not have been developed "specifically" to create a browser, it is a fair comment.
That said, Ladybird was started as part of the SerenityOS project. That entire project was built using C++. If the original goal of Serenity was to build an opeerating system, C++ would have felt like a reasonable choice at the time.
By the time Ladybird was looking for "better" languages than C++, Ladybird was already a large project and was making very heavy use of traditional OOP. Rust was evaluated but rejected because it did not support OOP well. Or, at least, it did not support integration into a large, C++ based, OOP project.
Perhaps, if Ladybird had first selected a languge to write a browser from scratch, they would have gone with Rust. We will never know,
We do know that Mozilla, despite being the de facto stewards of Rust at the time, and having a prototype web browser written in Rust (Servo), decided to drop both Rust and Servo. So, perhaps using Rust for browsers is not as open and shut as you imply.
I think you are conflating the development of Servo with the design and development of Rust.
20240810 https://news.ycombinator.com/item?id=41208836 Ladybird browser to start using Swift language this fall
- Excellent for short-lived programs that transform input A to output B
- Clunky for long-lived programs that maintain large complex object graphs
- Really impressive ecosystem
- Toxic community
Lots of people seem really committed to OOP. Rust is definitely a bad fit if you can't imagine writing code without classes and objects. I don't think this makes rust is a bad language for the problem. Its just, perhaps, makes rust a bad language for some programmers.
It isn't a _tracing_ GC
Every browser in use is stuck with C++ because they're in way too deep at this point, but Chromium and Firefox are both chipping away at it bit by bit and replacing it with safer alternatives where they feasibly can. Chromium even blocked JPEG-XL adoption until there was a safe implementation because they saw the reference C++ decoder as such a colossal liability.
IMO the takeaway is that although those browsers do use a ton of C++ and probably always will, their hard-won lessons have led them to wish they didn't have to, and to write a brand new browser in C++ is just asking to needlessly repeat all of the same mistakes. Chromium uses C++ because Webkit used C++ because KHTML used C++ in 1998. Today we have the benefit of hindsight.
Quickly followed by several vulnerabilities in that reference library as well; good move
And Andreas Kling already proved the naysayers wrong when he showd that a new operating system and Web browser can be written entirely from scratch, the former not even using any standard libraries; so beware when you are inclined to say 'not feasible'.
What is this mythical subset of C++? Does it include use of contemporary STL features like string_view? (Don’t get me wrong — modern STL is considerably improved, but it’s not even close to being memory-safe.)
Ladybird inherits its C++ from SerenityOS. Ladybird has an almost completely homegrown standard library including their own pointer classes and a couple of different string classes that do some interesting things with memory. But perhaps the most novel stuff are things like TRY and MUST: https://github.com/SerenityOS/serenity/blob/master/Documenta...
You see this reflected all the way back to the main function. Here is the main entry function for the entire browser:
ErrorOr<int> ladybird_main(Main::Arguments arguments).
https://github.com/LadybirdBrowser/ladybird/blob/master/UI/Q...
If Ladybird is successful, I would not be surprised to see its standard library take off with other projects. Again, it is really the SerenityOS standard library but the SerenityOS founder left the project to focus on Ladybird. So, that is where this stuff evolves now.
I can totally imagine how a prolific and ambitious developer would create a world of their own, essentially another language with domain-specific vocabulary and primitives. People often talk about using a "subset of C++" to make it manageable for mortals, and I think the somewhat unusual consideration of Swift was related to this desire for an ergonomic language to express and solve the needs of the project.
class FontFeatureValuesMapIterationSource final
: public PairSyncIterable<CSSFontFeatureValuesMap>::IterationSource {
public:
FontFeatureValuesMapIterationSource(const CSSFontFeatureValuesMap& map,
const FontFeatureAliases* aliases)
: map_(map), aliases_(aliases), iterator_(aliases->begin()) {}Ranges are not memory safe. Sorry.
Having a checklist of "things not to do" is historically a pretty in effectiveway to ensure memory safety, which is why the parent comment was asking for details. The fact that this type of thing gets dismissed as a non-issue is honestly a huge part of the problem in my opinion; it's time to move on from pretending this is a skill issue.
Moreover, Servo aims to be embeddable (there are some working examples already), which is where other non-Chrome/ium browsers are failing (and Firefox too). Thanks to this it has much better chance at wider adoption and actually spawning multiple browsers.
https://github.com/LadybirdBrowser/ladybird/tree/master/UI/A...
It's frustrating to discuss. It is a wonderful case study in how not to make engineering management decisions, and yet, they've occurred over enough time, and the cause is appealing enough, that it's hard to talk about out loud in toto without sounding like a dismissive jerk.
From what I can tell they're pretty laser focused on making a browser (even in this issue, they're abandoning Swift).
I agree with you. I also agree that this decision is an example of that.
SerenityOS had an "everything from scratch in one giant mono-repo" rule. It was, explicitly a hobby project and one rooted in enjoyment and 'idealism from the get go. It was founded by a man looking for something productive to focus on instead of drugs. It was therapy. Hence the name.
Ladybird, as an independent project, was founded with the goal of being the only truly independent web browser (independent from corporate control generally and Google specifically).
They have been very focussed on that, have not had any sacred cows, and have shed A LOT of the home-grown infrastructure they inherited from being part of SerenityOS. Sometimes that saddens me a little but there is no denying that it has sped them up.
Their progress has been incredible. This comment is being written in Ladybird. I have managed GitHub projects in Ladybird. I have sent Gmail messages in Ladybird. It is not "ready" but it blows my mind how close it is.
I think Ladybid will be a "usable" browser before we enter 2027. That is just plain amazing.
Swift is a poorly designed language, slow to compile, visibly not on path to be major system language, and they had no expert on the team.
I am glad they are cutting their losses.
Also funny enough, all cross platform work is with small work groups, some even looking for funding … anyway.
Apple has been always 'transactional' when it comes to OSS - they open source things only when it serves a strategic purpose. They open-sourced Swift only because they needed the community to build an ecosystem around their platform.
Yeah, well, sure they've done some work around LLVM/Clang, WebKit, CUPS, but it's really not proportional to the size and the influence they still have.
Compare them to Google, with - TensorFlow, k8s, Android (nominally), Golang, Chrome, and a long tail of other shit. Or Meta - PyTorch and the Llama model series. Or even Microsoft, which has dramatically reversed course from its "open source is a cancer" era (yeah, they were openly saying that, can you believe it?) to becoming one of the largest contributors on GitHub.
Apple I've heard even have harshest restrictions about it - some teams are just not permitted to contribute to OSS in any way. Obsessively secretive and for what price? No wonder that Apple's software products are just horrendously bad, if not all the time - well, too often. And on their own hardware too.
I wouldn't mind if Swift dies, I'm glad Objective-C is no longer relevant. In fact, I can't wait for Swift to die sooner.
Sort of an exception that proves the rule. Yes, it's great and was released for free. But at least partially that's not a strategic decision from Apple but just a requirement of the LGPLv2 license[1] under which they received it (as KHTML) originally.
And even then, it was Blink and not WebKit that ended up providing better value to the community.
[1] It does bear pointing out that lots of the new work is dual-licensed as 2-clause BSD also. Though no one is really trying to test a BSD-only WebKit derivative, as the resulting "Here's why this is not a derived work of the software's obvious ancestor" argument would be awfully dicey to try to defend. The Ship of Theseus is not a recognized legal principle, and clean rooms have historically been clean for a reason.
My understanding is that by default you are not allowed to contribute to open-source even if its your own project. Exceptions are made for teams whose function is to work on those open-source project e.g. Swift/LLVM/etc...
Apple is (was?) good at hardware design and UX, but they pretty bad at producing software.
I haven’t used it in a couple decades, but I do remember it fondly. I also suspect I’d hate it nowadays. Its roots are in a language that seemed revolutionary in the 80s and 90s - Smalltalk - and the melding of it with C also seemed revolutionary at the time. But the very same features that made it great then probably (just speculating - again I haven’t used it in a couple decades) aren’t so great now because a different evolutionary tree leapfrogged ahead of it. So most investment went into developing different solutions to the same problems, and ObjC, like Smalltalk, ends up being a weird anachronism that doesn’t play so nicely with modern tooling.
I haven’t used it in Apple’s ecosystem, so maybe I am way off base here. But it seems to me that it was Apple’s effort to evolve the language away from its systems roots into a more suitable applications language that caused all the ugliness.
My read is that that this is the main thing that happened here.
The Ladybird team is quite pragmatic. Or, at least their founder is. I think they understood the scale of buidling a browser. They understood that it was a massive undertaking. They also understood the security challenge of building an application so big that processes untrusted input as its main job. So, they thought that perhaps they needed a better language than C++ for these reasons. They evaluated a few options and came away thinking that Swift was the best, so they announced that they expected to move towards Swift in the future.
But the other side of being pragamatic is that, as time passed, they also realized how hard it would be to chagne horses. And they are quite productivity driven. They produce a YouTube video every month detailing their progress including charts and numbers.
Making no progress on the "real" job of building a browser to make progress on the somewhat artificial job of moving to a new programming language just never made sense I guess.
And the final part of being pragmatic is that, after months of not really making any progress on the switch, the founder posted this patch essentially admitting the reality and suggesting they just formalize the lack of progress and move on.
The lack of progress on Swift that is. Their progress on making a browser has been absolutely mind-blowing. This comment is being written in Ladybird.
And someone will be stuck not to do anything because they are unsatisfied with all languages. :-)
Edit: I explained my position better below.
LLVM: Pretty much everyone who has created a programming language with it has complained about its design. gingerbill, Jon Blow, and Andrew Kelley have all complained about it. LLVM is a good idea, but it that idea was executed better by Ken Thompson with his C compiler for Plan 9, and then again with his Go compiler design. Ken decided to create his own "architecture agnostic" assembly, which is very similar to the IR idea with LLVM.
Swift: I was very excited with the first release of Swift. But it ultimately did not have a very focused vision outlined for it. Because of this, it has morphed into a mess. It tries to be everything for everyone, like C++, and winds up being mediocre, and slow to compile to top it off.
Mojo isn't doesn't exist for the public yet. I hope it turns out to be awesome, but I'm just not going to get my hopes up this time.
LLVM is
- Slow to compile
- Breaks compilers/doesn't have a stable ABI
- Optimizes poorly (at least, worse than GCC)
Mojo... I have things I could say... But I'll stick to this. I talked to engineers there and I asked one how they expected any python developers to use the planned borrow checker. The engineer said "Don't worry about it" ie they didn't have a plan. The nicest thing I can say is they didn't bullshit me 100% of the time when I directly asked a question privately. That's the only nice or neutral thing I could say
I suggest you ask around to see what the consensus is for which compiler is actually mature. Hint: for all its warts, nobody is writing a seriously optimized language in any of the options you listed besides LLVM.
You could imagine creating a modified Go assembler that is more generic and not tied to Go's ABI that could accomplish the same effect as LLVM. However, it'd probably be better to create a project like that from scratch, because most of Go's optimizations happen before reaching the assembler stage.
It would probably be best to have the intermediate language that QBE has and transform that into "intermediate assembly" (IA) very similar to Go's assembly. That way the IL stage could contain nearly all the optimization passes, and the IA stage would focus on code generation that would translate to any OS/arch combo.
The latter is definitely a defining capability of Anders Hejlsberg. (C#/Typescript designer)
I don't have an emotional reaction to this, i.e. I don't think you're being mean, but it is wrong and reductive, which people usually will concisely, and perhaps reductively, describe as "mean".
Why is it wrong?
LLVM is great.
Chris Lattner left Apple a *decade* ago, & thus has ~0 impact or responsibility on Swift interop with C++ today.
Swift is a fun language to write, hence, why they shoehorned it in, in the first place.
Mojo is fine, but I wouldn't really know how you or I would judge it. For me, I'm not super-opinionated on Python, and it doesn't diverge heavily from it afaik.
I was hired around Google around the same time, but not nearly as famous :)
AFAICouldT it was a "hire first, figure out what to do later", and it ended up being Swift for TensorFlow. That went ~nowhere, and he left within 2 years.
That's fine and doesn't reflect on him, in general, that's Google for ya. At least that era of Google.
If he's such a horrible engineer then we should have lots of LLVM replacements, right?
It's a cool project and I'd consider it for a toy language but it's far from an LLVM replacement.
What are other projects trying something similar that deserve attention?
But you should perhaps give your attention to Servo. They were founded as a project to write a modern browser in Rust. So, no hand-waving there.
No hand-waving on the Ladybird team either in my opinion. They have very strong technical leadership. The idea that building a massive application designed to process untrusted user input at scale might need a better language than C++ seems like a pretty solid technical suggestion. Making incedible progress month after month using the language you started with seems pretty good too. And deciding, given the progress buidling and the lack of progress exploring the new language, that perhaps it would be best to formally abandon the idea of a language switch...well, that seems like a pretty solid decision as well.
At least, that is my view.
Oh, and I was a massive Servo fan before the Ladybird project even began. But, given how much further Ladybird has gotten than Servo has, despite being at it for less time and taking on a larger scope...well, I am giving my attention to Ladybird these days.
This comment was written in Ladybird.
Carefully making decisions and then reassessing those choices later on when they prove to be problematic is the opposite of handwavey...