Conversation

Notices

1. Embed this notice
   lambdageek@mastodon.social's status on Saturday, 23-Dec-2023 09:41:47 JST lambdageek

   • Haelwenn /элвэн/ :triskell:
   • niconiconi
   • Joe

   @niconiconi @lanodan @not2b
   > shell scripts are inherently sequential

   Funny you should say that...

   https://www.usenix.org/conference/osdi22/presentation/kallas

   This research group wrote a JIT(!) for bash(!!) that also automatically parallelizes the code (!!!)

   • Embed this notice
     Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Saturday, 23-Dec-2023 09:41:46 JST Haelwenn /элвэн/ :triskell:

     in reply to

     • niconiconi
     • Joe

     @lambdageek @niconiconi @not2b Sounds like the best way to have a ton of bugs everywhere if you'd use it as /bin/sh

   • Embed this notice
     Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Saturday, 23-Dec-2023 10:31:13 JST Haelwenn /элвэн/ :triskell:

     in reply to

     • niconiconi
     • dpflug
     • Joe
     @dpflug @lambdageek @niconiconi @not2b Jit for an init?
     
     https://www.tuhs.org/cgi-bin/utree.pl?file=SysIII/etc/rc obviously doesn't needs that.
   • Embed this notice
     dpflug (dpflug@hachyderm.io)'s status on Saturday, 23-Dec-2023 10:31:14 JST dpflug

     in reply to

     • Haelwenn /элвэн/ :triskell:
     • niconiconi
     • Joe

     @lambdageek
     :blobfoxhyper: Let's built an init with it
     @lanodan @niconiconi @not2b

   • Embed this notice
     lambdageek@mastodon.social's status on Saturday, 23-Dec-2023 10:31:15 JST lambdageek

     in reply to

     • Haelwenn /элвэн/ :triskell:
     • niconiconi
     • Joe

     @lanodan @niconiconi @not2b

     From the abstract:
     > When run on a wide variety of benchmarks, including the POSIX shell test suite, PaSh-JIT (1) does not break scripts, even in cases that are likely to break shells in widespread use; and (2) offers significant speedups, whenever parallelization is possible. These results show that PaSh-JIT can be used as a drop-in replacement for any non-interactive shell use, providing significant speedups without any risk of breakage.

     So... no?

   • Embed this notice
     Oblomov (oblomov@sociale.network)'s status on Saturday, 23-Dec-2023 16:23:28 JST Oblomov

     in reply to

     • Haelwenn /элвэн/ :triskell:
     • niconiconi
     • Joe

     @lambdageek @niconiconi @lanodan @not2b

     I'm scared.

     Haelwenn /элвэн/ :triskell: likes this.