Conversation

Notices

1. Embed this notice
   Christine Lemmer-Webber (cwebber@social.coop)'s status on Saturday, 21-Jun-2025 08:56:41 JST Christine Lemmer-Webber

   The Little Learner and hotel room hacking https://dustycloud.org/blog/the-little-learner-and-hotel-room-hacking/

   Everyone is posting their AI opinions lately and hey, I want to post mine too. But I want to be well informed, and understand things on an *implementation* level. Some thoughts about the experience of going through a lovely book that's helping me do that while I am ostensibly on vacation!

   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Saturday, 21-Jun-2025 14:25:09 JST Christine Lemmer-Webber

     in reply to

     • Shriram Krishnamurthi

     @shriramk It's really strange because I've used Racket and DrRacket plenty! But not on ARM64 I guess? Startup time was around five to ten minutes, which is strange, or at least it was after I installed the malt package (not sure what that would have to do with it though)

   • Embed this notice
     Shriram Krishnamurthi (shriramk@mastodon.social)'s status on Saturday, 21-Jun-2025 14:25:10 JST Shriram Krishnamurthi

     in reply to

     @cwebber Yeah, I enjoyed the book! Too bad the Racket version didn't work fast enough for you, but glad you got that working.

   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Saturday, 21-Jun-2025 14:25:34 JST Christine Lemmer-Webber

     in reply to

     • Shriram Krishnamurthi

     @shriramk And I noticed your name in the acknowledgements! :)

   • Embed this notice
     Humberto Ortiz-Zuazaga (humberto@mastodon.online)'s status on Sunday, 22-Jun-2025 01:15:41 JST Humberto Ortiz-Zuazaga

     in reply to

     @cwebber

     I submitted a bug to debian on this. Startup on aarch64 jumped from a few seconds to several minutes on my MNT Reform 2. The guix version works fine.

     ```
     $ which racket
     /home/humberto/.guix-profile/bin/racket
     $ time racket -e "(* 5 6)"
     30

     real 0m3.246s
     ...
     $ time /usr/bin/racket -e "(* 5 6)"
     30

     real 3m39.905s
     ```

     https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1085959

   • Embed this notice
     David Bremner (bremner@mathstodon.xyz)'s status on Sunday, 22-Jun-2025 02:53:39 JST David Bremner

     in reply to

     • Humberto Ortiz-Zuazaga

     @humberto @cwebber Well that's embarrassing. The file /etc/racket/config.rktd is actually architecture dependent, but we've been shipping the amd64 version on all architectures in the package racket-common. I'll have to figure out the right long term fix, but for now a search and replace of x86_64 with aarch64 (two places) should fix it. If not, please follow up on the Debian bug, or here.

     Christine Lemmer-Webber repeated this.
   • Embed this notice
     David Bremner (bremner@mathstodon.xyz)'s status on Sunday, 22-Jun-2025 02:53:40 JST David Bremner

     in reply to

     • Humberto Ortiz-Zuazaga

     @humberto @cwebber Oh! I finally have some aarch64 hardware so I'll try to revisit that. I think when I looked at it before the difference was not as dramatic on the Debian porter boxes.

     Christine Lemmer-Webber repeated this.
   • Embed this notice
     Humberto Ortiz-Zuazaga (humberto@mastodon.online)'s status on Sunday, 22-Jun-2025 02:54:02 JST Humberto Ortiz-Zuazaga

     in reply to

     • David Bremner

     @bremner @cwebber Awesome! Times are now comparable with guix

     $ time /usr/bin/racket -e "(exit)"

     real 0m2.976s

     I guess we were running emulated x86_64 code.

     Haelwenn /элвэн/ :triskell: likes this.
     Christine Lemmer-Webber repeated this.
   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Sunday, 22-Jun-2025 02:54:57 JST Christine Lemmer-Webber

     in reply to

     • David Bremner
     • Humberto Ortiz-Zuazaga

     @humberto @bremner Wow, that does explain it! Dang!

     It's nice to hear this. I was feeling kind of depressed this morning, and feeling kind of useless, but it's good to know that just writing a little blogpost about playing with something and running into a surprising issue can uncover things that were indeed issues, even when you're just musing out loud.

     Haelwenn /элвэн/ :triskell: likes this.
   • Embed this notice
     David Bremner (bremner@mathstodon.xyz)'s status on Monday, 23-Jun-2025 07:58:17 JST David Bremner

     in reply to

     • Humberto Ortiz-Zuazaga

     @humberto @cwebber Bear with me a second, but it was actually impressively fast before, considering ... it was compiling the entire system from .rkt to bytecode (.zo) on every startup. That's a lot of compilation, and I think the Racket byte code compiler does a lot more than something like Python.

   • Embed this notice
     Seth "Twylo" Morabito (twylo@mastodon.sdf.org)'s status on Monday, 23-Jun-2025 07:58:25 JST Seth "Twylo" Morabito

     in reply to

     • David Bremner
     • Humberto Ortiz-Zuazaga

     @bremner @humberto @cwebber Just when I was feeling really burned out on open source you folks go and demonstrate positive results from open discussion and make me hopeful again. Good job!