Steel Bank Common Lisp version 2.6.7
sbcl.org123 points by tmtvl 3 hours ago
123 points by tmtvl 3 hours ago
I wonder sometimes how the world would look like if lisp won and the unit of deployment was a lisp machine image, if that makes sense. How would a kubernetes optimized for lisp look like? How would AWS EC2 look like? etc.
That (deployment is via a customized image) is the thing I find most awkward.
Somewhere, I have a copy of a commercial LISP for Windows which would compile to an executable --- apparently this sort of thing is still possible, but it's not widely known/used, and sadly Jean-Marie Hullot's "SOS Interface" for the Mac was co-opted to NeXTstep:
https://denninginstitute.com/itcore/userinterface/GUIHistory...
I'd dearly love to see a RAD (Rapid Application Development) tool using LISP w/ a nifty UI for working up a GUI which would compile to something easily deployed (maybe HTML5 Canvas and JavaScript) as a stand-alone/single-file web application?
Docker deploys is a bit like that I think. Overall, I think many languages and programming systems pick up pieces from each other nowadays. PHP almost by accident can be quite stateless. Yet it's looking more like Java. Many things which could be deployed as an executable, are deployed as VM images. I'm sure we can come up with more things...
We wouldn't have need such barbaric things in a lisp world. :p
Personally I think/hope they would have earlier discoveries / more broad usage of the deterministic systems like nix etc, which are built upon functional principles of immutability etc.
Maybe the world would be guix/Hurd!
Common lisp isn't functional in that sense.
Sure the language might not be immutable but the lisp machine alt universe's "overton window" would I imagine be more functional then today's (JVM? maybe that's the qnalog?) reality.
Common Lisp is pretty much the opposite of what people think of as functional nowadays. You're constantly changing the image and everything is mutable.
I would presume there was a lot of mixing of imperative/functional in the 80s where this alt universe starts. I think the same type of world in which lisp machines would have become king would also be similar
> Personally I think/hope they would have earlier discoveries / more broad usage of the deterministic systems like nix etc, which are built upon functional principles of immutability etc.
Yeah, that, uh, doesn't sound like Lisp
Not that different.
AWS already has lots of AMI and docker are essentially image based.
We would version whole images in something like git, and that's all.
Lisp's been around for 70 years at this point and hordes of developers tried it and said: lovely, but I ain't using it at work.
I'm strongly convinced, having used Scheme, CL, Racket, Clojure that Lisp is doomed to be (mostly) a hobby language.
The very power of Lisp, macros, dynamic programming, reflection lead to a mess of an ecosystem where every single developer reinvents the wheel and nobody can understand yet another DSL invented by the next developer next to them.
Racket's theme of being a "programming language for building programming languages" is just the poster child of this naivety: I don't want even more friction.
What scales and works is simple and boring.
That's by the way why also Haskell has struggled forever. Beyond its dreadful DX, poor tooling and unacceptable compilation times, the language is just plagued by compiler extensions and every single developer reinventing its own abstractions.
There was the simple haskell movement to just standardize around a set of extensions and conventions, but nope, these languages unavoidably attract people that want to stay in the ivory tower.
Really, I love both lisps and haskell, but I would take a dreadful php over them every single day at work. No contest.
The problem is that the type of developer you're talking about is the kind who invents a thousand excuses about anything. It doesn't begin and end with different languages. Using a different subset of their favorite language? A different toolchain? A radically different build process? You'll hear plenty of whining about it. Even just moving them to a different codebase will make them miserable. You put a Lisp or a Prolog in front of them, and their attention flows upwards to the most general principles to whine about. The entire thing is that if you stick them into an unfamiliar environment, they will vocalize their discomfort through rationalization. I think most people who have led teams will know exactly what I'm talking about.
Not that these people are necessarily bad at their jobs, but they're definitely a personality type you should learn to identify so you can manage them properly.
I disagree with this profoundly. There is no argument that Lisp in general can get you in a terrible mess. So can C, or pretty much any language (I well remember a case of 11-way multiple inheritance in C++ game code). If I were running a Lisp-based project, there would be project standards (like no exported macros without signoff, and a list of standard library dependencies), documentation requirements, and code review.
The fact that Lisp is the “programmable programming language” doesn't mean that every engineer should be inventing weird versions of while loops. What it does mean is that a skilled Lisp programmer can build a domain-specific language that substantially helps in development.
One good example of this is the Crash Bandicoot games, along with the same studio's Jax and Daxter, which were built with dialects of Lisp (yes, they had to compile the code, and take care with storage allocation, just like any other game code). Cisco hired Kent Dybvig, principal author of the Chez Scheme system, and open-sourced that software; I have no clue what they use it for, to be honest, but I assume that they had some reason for doing this. Companies using various Lisp languages have been documented in areas from fintech to quantum computing.
>> said: lovely, but I ain't using it at work.
Says who, you? Wrong. I use Common Lisp at work.
edit: and not just use as in a side toy, design and writing software in CL is my primary responsibility. FWIW, at a FAANG!
>FWIW, at a FAANG!
Do you work on ITA? Only FAANG I'm aware of using Lisp is Google.
Nope not on ITA. I work at the intersection of HW and SW. Lisp is valued for its strengths and I haven't come across any tool better to explore complex problem space. Of course, the opportunities are fewer, get filled up organically and are better suited for someone with valuable domain knowledge who can translate it into working space for a wider team using Lisp as the tool to write the spec!
> The very power of Lisp, macros, dynamic programming, reflection lead to a mess of an ecosystem where every single developer reinvents the wheel and nobody can understand yet another DSL invented by the next developer next to them.
Have you like looked at anything in the Clojure ecosystem because this description makes no sense of any kind. Clojure has macros but less powerful than common lisp and Clojure code in the ecosystem tends to be small and understandable.
It seems to me the Clojure community internalized the lesson of not writing macros unless it's really necessary more than other Lisps did.
If any SBCL dev is here, please add documentation for how to use the memory arena feature. The only doc is an very old proposal document.
Yeah I'm not sure why it's not in the manual, as it's had arena allocation support since at least 2.4.x, but basically:
- use SB-VM:NEW-ARENA to make a new arena
- use SB-VM:WITH-ARENA to redirect ordinary allocation into an existing arena like you would use WITH-OPEN-FILE or similar macros
The only real doc is this internals note, and it doesn't even cover NEW-ARENA which I guess is left as an exercise to the reader: https://github.com/sbcl/sbcl/blob/master/doc/internals-notes...
* the SB-SIMD contrib now supports ARM64. (Thanks to Sylvia Harrington)
* AVX512 instructions are now supported on X86-64. (Thanks to Robert Smith and Arthur Miller)
* additional support for SIMD instructions on ARM64 and X86-64. (Thanks to Arthur Miller)
No auto-vectorization yet, but you may find this illuminating: https://www.stylewarning.com/posts/nbody
If memory serves, traditionally CCL (Clozure Common Lisp [0]) had better support for Windows but SBCL was unbeatable for speed. Is that still the case?
Great to see the project is still going strong, I kinda want to try CL but I always feel like I don't have a great use-case.
CL can be used pretty much anywhere nowadays, even on the web with ECL wasm compilation.
Anywhere is where exactly? I've been thinking of where I could use it for like 10 minutes, and I couldn't come up with anything. Maybe a game engine or web services. It wouldn't make much sense for any kind of desktop software or command line utilities that are supposed to start really fast and have minimal runtime.
A binary has a startup time of like 20 milliseconds, the overwhelming majority of that is setting up page tables iirc. While that's a lot for a small cli utility called in a loop, it's really nothing for a desktop application.
I wish more desktop applications could hit even a 100ms startup time. These days it feels like 5 seconds or more is the norm.
I have a todo list longer than the Tour de France I want to tackle first!
I'm going pretty deep on Lisp on accident - can anybody give me more context on this?
Great semver number