Conversation

Notices

1. Embed this notice
   niconiconi (niconiconi@mk.absturztau.be)'s status on Friday, 27-Jan-2023 09:23:43 JST niconiconi

   OpenVMS implies the existence of 386VMS, NetVMS and FreeVMS.

   • Embed this notice
     Roy Tam (roytam1@miniwa.moe)'s status on Friday, 27-Jan-2023 09:23:28 JST Roy Tam

     in reply to

     • wakame
     • chjara
     • La Ursidinoj (Bjornsdottirs)
     @wakame @chjara @ellenor2000 @niconiconi OpenDOS exists.
   • Embed this notice
     wakame (wakame@tech.lgbt)'s status on Friday, 27-Jan-2023 09:23:29 JST wakame

     in reply to

     • chjara
     • La Ursidinoj (Bjornsdottirs)

     @chjara @ellenor2000 @niconiconi
     I wonder if an SSL lib for FreeDOS exists, that would match the mood.

     Btw: FreeDOS implies the existence of OpenDOS, NetDOS and 386DOS.

   • Embed this notice
     chjara (chjara@snowdin.town)'s status on Friday, 27-Jan-2023 09:23:31 JST chjara

     in reply to

     • La Ursidinoj (Bjornsdottirs)
     @niconiconi @ellenor2000 i'm sure a naive C implementation with a good optimizing compiler would have more than satisfying performance for that
   • Embed this notice
     niconiconi (niconiconi@mk.absturztau.be)'s status on Friday, 27-Jan-2023 09:23:33 JST niconiconi

     in reply to

     • La Ursidinoj (Bjornsdottirs)

     @ellenor2000@mastodon.top 386SSL sounds like a cool would-be retrocomputing project. Hand-tuned i386 assembly for X25519 and ChaCha20 to browse the modern web.

   • Embed this notice
     La Ursidinoj (Bjornsdottirs) (ellenor2000@mastodon.top)'s status on Friday, 27-Jan-2023 09:23:41 JST La Ursidinoj (Bjornsdottirs)

     in reply to

     @niconiconi No it doesn't, no more than NetSSL, 386SSL, or FreeSSL are implied by OpenSSL.

     There is a FreeVMS in the wild.

   • Embed this notice
     Roy Tam (roytam1@miniwa.moe)'s status on Friday, 03-Feb-2023 18:15:33 JST Roy Tam

     in reply to

     • La Ursidinoj (Bjornsdottirs)
     @niconiconi @ellenor2000 TLS 1.2 TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305_SHA256 (0xcca9) is working with NSS-3.31.4-Retrozilla on 386DX here.
     and you don't even need hand-writing assembly.
     
   • Embed this notice
     Roy Tam (roytam1@miniwa.moe)'s status on Friday, 03-Feb-2023 18:17:29 JST Roy Tam

     in reply to

     • Roy Tam
     • La Ursidinoj (Bjornsdottirs)
     @niconiconi @ellenor2000 s/31/21/;
     typed and posted too fast without checking.
   • Embed this notice
     Roy Tam (roytam1@miniwa.moe)'s status on Friday, 03-Feb-2023 18:53:47 JST Roy Tam

     in reply to

     • La Ursidinoj (Bjornsdottirs)
     @niconiconi @ellenor2000 oh, my NSS-3.21.4-Retrozilla doesn't do x25519.
   • Embed this notice
     niconiconi (niconiconi@mk.absturztau.be)'s status on Friday, 03-Feb-2023 18:53:48 JST niconiconi

     in reply to

     • Roy Tam
     • La Ursidinoj (Bjornsdottirs)

     @roytam1@miniwa.moe @ellenor2000@mastodon.top Correction: I meant X25519. ChaCha20 has no bignum multiplication...

   • Embed this notice
     niconiconi (niconiconi@mk.absturztau.be)'s status on Friday, 03-Feb-2023 18:53:50 JST niconiconi

     in reply to

     • Roy Tam
     • La Ursidinoj (Bjornsdottirs)

     @roytam1@miniwa.moe @ellenor2000@mastodon.top Hand-tuning can potentially make it faster. There's a large number of published literature on optimizing ChaCha20 for embedded systems that can potentially be applied back to the 386. I remember reading about an optimized version of Karatsuba multiplication with the least number of steps.