home.social

#sbcl — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #sbcl, aggregated by home.social.

  1. @djl So far I have learned and experimented with only. Though I used it preliminary on Debian, I installed it on Windows also to make binaries for Window$

    How about . I will try it when I use and I think my learning of Lisp would be useful for many dialects, right?

    Btw, I released my first code under

  2. @djl So far I have learned and experimented with #SBCL only. Though I used it preliminary on Debian, I installed it on Windows also to make binaries for Window$

    How about #GNU #Guile. I will try it when I use #Guix and I think my learning of Lisp would be useful for many dialects, right?

    Btw, I released my first code under #AGPLV3

  3. @djl So far I have learned and experimented with #SBCL only. Though I used it preliminary on Debian, I installed it on Windows also to make binaries for Window$

    How about #GNU #Guile. I will try it when I use #Guix and I think my learning of Lisp would be useful for many dialects, right?

    Btw, I released my first code under #AGPLV3

  4. @djl So far I have learned and experimented with #SBCL only. Though I used it preliminary on Debian, I installed it on Windows also to make binaries for Window$

    How about #GNU #Guile. I will try it when I use #Guix and I think my learning of Lisp would be useful for many dialects, right?

    Btw, I released my first code under #AGPLV3

  5. @djl So far I have learned and experimented with #SBCL only. Though I used it preliminary on Debian, I installed it on Windows also to make binaries for Window$

    How about #GNU #Guile. I will try it when I use #Guix and I think my learning of Lisp would be useful for many dialects, right?

    Btw, I released my first code under #AGPLV3

  6. 🚀 Ah, the riveting world of #SBCL #assembly code, where 5 whole bytes are shaved off for a more *efficient* space-wasting experience. Paul Khuong and friends #debate byte-lengths like they're splitting atoms, because clearly, optimizing a few machine instructions is the apex of #programming 🧠✨.
    pvk.ca/Blog/2014/03/15/sbcl-th #byte-optimization #efficient #machine-instructions #HackerNews #ngated

  7. 🚀 Ah, the riveting world of #SBCL #assembly code, where 5 whole bytes are shaved off for a more *efficient* space-wasting experience. Paul Khuong and friends #debate byte-lengths like they're splitting atoms, because clearly, optimizing a few machine instructions is the apex of #programming 🧠✨.
    pvk.ca/Blog/2014/03/15/sbcl-th #byte-optimization #efficient #machine-instructions #HackerNews #ngated

  8. I've been a #DoomEmacs user for some time, now, even having curated my own literate config out of their "static" files - along with all the extra stuff one adds to such a thing.

    As one does, in the past I've also attempted to do my own #emacs config from scratch, usually falling into the "let's recreate doom!" hole, whether I wanted to or not (thus the reason I just went the route I did).

    Over the past couple of months I've decided to - slowly - do the emacs bankruptcy declaration. Now, just to make things even more interesting I have decided against the use-package macro. Why? Because why not!

    I've been using my scratch build for a month or so, haven't even opened my doom config up. It's been a great process, learning process, more understand of the process. It helps that I've been on a learn lisp quest for the past couple of years, too (#CommonLisp, #sbcl, #clojure, #elisp of course).

    The large emacs distros are great an an excellent way to find curated packages. Doom is even a really nice framework. Sometimes you just want to do your own thing.

  9. I've been a #DoomEmacs user for some time, now, even having curated my own literate config out of their "static" files - along with all the extra stuff one adds to such a thing.

    As one does, in the past I've also attempted to do my own #emacs config from scratch, usually falling into the "let's recreate doom!" hole, whether I wanted to or not (thus the reason I just went the route I did).

    Over the past couple of months I've decided to - slowly - do the emacs bankruptcy declaration. Now, just to make things even more interesting I have decided against the use-package macro. Why? Because why not!

    I've been using my scratch build for a month or so, haven't even opened my doom config up. It's been a great process, learning process, more understand of the process. It helps that I've been on a learn lisp quest for the past couple of years, too (#CommonLisp, #sbcl, #clojure, #elisp of course).

    The large emacs distros are great an an excellent way to find curated packages. Doom is even a really nice framework. Sometimes you just want to do your own thing.

  10. I updated my "is #guile fast?" page with the new r7rs #benchmarks data and anchored Guile and #Chez in the benchmarksgame:

    #Scheme benchmarks: draketo.de/software/guile-fast
    Compared to other languages: draketo.de/software/guile-fast

    Chez is roughly factor 7 slower than C,
    Guile roughly factor 18
    (#Java 3.5, #SBCL 5, #Racket 12, #Ruby 60 and #Python 70)

  11. I updated my "is #guile fast?" page with the new r7rs #benchmarks data and anchored Guile and #Chez in the benchmarksgame:

    #Scheme benchmarks: draketo.de/software/guile-fast
    Compared to other languages: draketo.de/software/guile-fast

    Chez is roughly factor 7 slower than C,
    Guile roughly factor 18
    (#Java 3.5, #SBCL 5, #Racket 12, #Ruby 60 and #Python 70)

  12. The operating system is a Debian trixie derivative. Out of the box, it has Firefox 131.0.2 and Xfce 4.18. apt-get install build-essential gave me gcc-14; other items in the repos: nodejs 20.17.0, emacs 29.4, golang 1.23. Elixir's still at 1.14.0 (aside: erlang won't get a JIT for RISC-V any time soon), Python's at 3.12, rustc 1.80.1 (but of course you can use rustup).

    Anyway, on to some benchmarks. glmark2-es2 reports a score of 1714, which is surprisingly 38% higher than @geerlingguy 's benchmark of the HiFive Premier P550.

    My real test is compiling sbcl; it's not in Debian or Ubuntu's repositories for RISC-V. I bootstrap it with GNU CLISP, and then rebuild it with itself, with sh ./make.sh --with-sb-doc --without-sb-thread. Unfortunately, I believe the build is single-core; I'm not sure if it's possible to use all the cores on my system for it.

    Times to rebuild sbcl with itself, including modules:

    Lichee Pi 3A (Ubuntu 24.04 derivative): 30 minutes
    VisionFive 2 (Ubuntu 24.04):            20 minutes
    Megrez (Debian trixie derivative):      12 minutes
    Ryzen 9900x (Ubuntu 24.04):              1 minute
    

    So, progress, but a long way to go.

    (Incidentally, both the Megrez and my desktop have 6400 memory.)

    #milkv #megrez #riscv #sbcl

  13. The operating system is a Debian trixie derivative. Out of the box, it has Firefox 131.0.2 and Xfce 4.18. apt-get install build-essential gave me gcc-14; other items in the repos: nodejs 20.17.0, emacs 29.4, golang 1.23. Elixir's still at 1.14.0 (aside: erlang won't get a JIT for RISC-V any time soon), Python's at 3.12, rustc 1.80.1 (but of course you can use rustup).

    Anyway, on to some benchmarks. glmark2-es2 reports a score of 1714, which is surprisingly 38% higher than @geerlingguy 's benchmark of the HiFive Premier P550.

    My real test is compiling sbcl; it's not in Debian or Ubuntu's repositories for RISC-V. I bootstrap it with GNU CLISP, and then rebuild it with itself, with sh ./make.sh --with-sb-doc --without-sb-thread. Unfortunately, I believe the build is single-core; I'm not sure if it's possible to use all the cores on my system for it.

    Times to rebuild sbcl with itself, including modules:

    Lichee Pi 3A (Ubuntu 24.04 derivative): 30 minutes
    VisionFive 2 (Ubuntu 24.04):            20 minutes
    Megrez (Debian trixie derivative):      12 minutes
    Ryzen 9900x (Ubuntu 24.04):              1 minute
    

    So, progress, but a long way to go.

    (Incidentally, both the Megrez and my desktop have 6400 memory.)

    #milkv #megrez #riscv #sbcl

  14. The operating system is a Debian trixie derivative. Out of the box, it has Firefox 131.0.2 and Xfce 4.18. apt-get install build-essential gave me gcc-14; other items in the repos: nodejs 20.17.0, emacs 29.4, golang 1.23. Elixir's still at 1.14.0 (aside: erlang won't get a JIT for RISC-V any time soon), Python's at 3.12, rustc 1.80.1 (but of course you can use rustup).

    Anyway, on to some benchmarks. glmark2-es2 reports a score of 1714, which is surprisingly 38% higher than @geerlingguy 's benchmark of the HiFive Premier P550.

    My real test is compiling sbcl; it's not in Debian or Ubuntu's repositories for RISC-V. I bootstrap it with GNU CLISP, and then rebuild it with itself, with sh ./make.sh --with-sb-doc --without-sb-thread. Unfortunately, I believe the build is single-core; I'm not sure if it's possible to use all the cores on my system for it.

    Times to rebuild sbcl with itself, including modules:

    Lichee Pi 3A (Ubuntu 24.04 derivative): 30 minutes
    VisionFive 2 (Ubuntu 24.04):            20 minutes
    Megrez (Debian trixie derivative):      12 minutes
    Ryzen 9900x (Ubuntu 24.04):              1 minute
    

    So, progress, but a long way to go.

    (Incidentally, both the Megrez and my desktop have 6400 memory.)

    #milkv #megrez #riscv #sbcl

  15. The operating system is a Debian trixie derivative. Out of the box, it has Firefox 131.0.2 and Xfce 4.18. apt-get install build-essential gave me gcc-14; other items in the repos: nodejs 20.17.0, emacs 29.4, golang 1.23. Elixir's still at 1.14.0 (aside: erlang won't get a JIT for RISC-V any time soon), Python's at 3.12, rustc 1.80.1 (but of course you can use rustup).

    Anyway, on to some benchmarks. glmark2-es2 reports a score of 1714, which is surprisingly 38% higher than @geerlingguy 's benchmark of the HiFive Premier P550.

    My real test is compiling sbcl; it's not in Debian or Ubuntu's repositories for RISC-V. I bootstrap it with GNU CLISP, and then rebuild it with itself, with sh ./make.sh --with-sb-doc --without-sb-thread. Unfortunately, I believe the build is single-core; I'm not sure if it's possible to use all the cores on my system for it.

    Times to rebuild sbcl with itself, including modules:

    Lichee Pi 3A (Ubuntu 24.04 derivative): 30 minutes
    VisionFive 2 (Ubuntu 24.04):            20 minutes
    Megrez (Debian trixie derivative):      12 minutes
    Ryzen 9900x (Ubuntu 24.04):              1 minute
    

    So, progress, but a long way to go.

    (Incidentally, both the Megrez and my desktop have 6400 memory.)

    #milkv #megrez #riscv #sbcl

  16. Having been informed of the Autumn Lisp Game Jam coming up: how's #CommonLisp GC performance for games nowadays? Last I checked, it seemed like all the major libre CL impls (mainly #SBCL & #Clozure) went full stop-the-world, which feels like a recipe for latency spikes. I saw flashes of stuff on sbcl-devel a while back about someone working on a lower-latency GC, but I don't know whether it got anywhere.

  17. #sbcl #lisp #commonLisp #asdf #executable #gui #build #tutorial #clim #mcclim #example I wrote for @theruran a while ago.

    You could basically submit this with literally any modification as a game for the #springlispgamejam2024 and it will be pretty similar to mine. I think it's a great approach to featureful interactive user interfaces for games.

    lispy-gopher-show.itch.io/logo

    itch.io/jam/spring-lisp-game-j

  18. Oh, I'm this 🤏 close to finishing up my work to convert #SBCL test coverage reports (`sb-cover:report`) to a format understandable by #Codecov 🕺

    That would mean, to say the least, I can have test coverage report badges for my #CommonLisp projects on Github 🤤

    Stay tuned 📻

  19. Oh, I'm this 🤏 close to finishing up my work to convert #SBCL test coverage reports (`sb-cover:report`) to a format understandable by #Codecov 🕺

    That would mean, to say the least, I can have test coverage report badges for my #CommonLisp projects on Github 🤤

    Stay tuned 📻

  20. Released an updated version (0.3.1) of sijo-doctest, a #doctest library for
    #CommonLisp.

    This includes some quality of life features, bugfixes and support for #SBCL, #ECL, #CCL and #ABCL.

    github.com/simendsjo/sijo-doct

  21. Released an updated version (0.3.1) of sijo-doctest, a library for
    .

    This includes some quality of life features, bugfixes and support for , , and .

    github.com/simendsjo/sijo-doct

  22. Released an updated version (0.3.1) of sijo-doctest, a #doctest library for
    #CommonLisp.

    This includes some quality of life features, bugfixes and support for #SBCL, #ECL, #CCL and #ABCL.

    github.com/simendsjo/sijo-doct

  23. Today I also officially forked #SBCL so I can transform my constant complaining into contributing. Officially.

    Might as well put #GitHub #CopilotChat to work for the whole #CommonLisp community, right?

    #GitHubCopilot #GitHubCopilotChat #GenerativeAI #Lisp #OpenSource

  24. Today I also officially forked #SBCL so I can transform my constant complaining into contributing. Officially.

    Might as well put #GitHub #CopilotChat to work for the whole #CommonLisp community, right?

    #GitHubCopilot #GitHubCopilotChat #GenerativeAI #Lisp #OpenSource