Conversation

Notices

1. Nate Cull (natecull@mastodon.social)'s status on Friday, 10-Jun-2022 04:39:41 UTC Nate Cull

   Fundamental Abstractions From Computing In 2022 That Might Still Exist And Be A Reliable Means Of Data Retrieval In 2122:

   1. Bytes (or at least integers from 0 to 255)
   2. Arrays (of things that could include either bytes, or integers from 0 to 255, and with a separate "length" integer)
   3. UTF-8 (or at least the 7-bit ASCII-like subset of it)
   4. .... ...

   Any takers for whether there even will be a 4?

   • Nate Cull (natecull@mastodon.social)'s status on Friday, 10-Jun-2022 04:39:41 UTC Nate Cull

     in reply to

     "I do not know with what Software-as-a-Service Platform War 3 will be fought, but Platform War 4 will be fought with pigeons and clay tablets"

     -- Einstein (Fred Einstein of Einstein 8-Track & Betamax Repairs, Mudville, Oklahoma, 1953)

     Santa Claes 🇸🇪🇭🇰🎅 likes this.
   • Nate Cull (natecull@mastodon.social)'s status on Friday, 10-Jun-2022 07:15:53 UTC Nate Cull

     in reply to

     • Adrian Cochrane

     @alcinnz

     We'll get Danny Hillis to etch a few copies of this year's Unicode database into some cogwheel in the Long Now clock.

     He might get a bit annoyed if we keep coming back for more cogs each year as the Unicode standard updates though...

     Santa Claes 🇸🇪🇭🇰🎅 likes this.
   • Adrian Cochrane (alcinnz@floss.social)'s status on Friday, 10-Jun-2022 07:15:54 UTC Adrian Cochrane

     in reply to

     @natecull Do make sure to print out the Unicode standard & store it in a library somewhere! Ideally as many libraries we can reach!

     I don't want to limit the, say, Chinese to ASCII...

   • Paolo Redaelli (paoloredaelli@mastodon.uno)'s status on Saturday, 11-Jun-2022 17:03:30 UTC Paolo Redaelli

     in reply to

     @natecull #ocaml, #rust and #Haskell all have specific packages in #Fedora and #Debian to deal with #endianness so it seems to matter. Also add that according to utfcheck Debian package description UTF-16 can be encoded in big or little-endian

   • Nate Cull (natecull@mastodon.social)'s status on Saturday, 11-Jun-2022 17:03:30 UTC Nate Cull

     in reply to

     • Paolo Redaelli

     @paoloredaelli

     Yep, UTF-16 has endianness and that's one of the many reasons why I think it's UTF-8, not UTF-16, that's a viable candidate for surviving the next hundred years.

     Santa Claes 🇸🇪🇭🇰🎅 likes this.
   • Nate Cull (natecull@mastodon.social)'s status on Saturday, 11-Jun-2022 17:03:31 UTC Nate Cull

     in reply to

     • Paolo Redaelli

     @paoloredaelli

     Endianness isn't really a thing in wire encodings or bytecode anymore, though, I think? Only in raw platform-specific RAM dumps of machine words, which we would generally want to avoid for security reasons, as well as portability.

     If we're very lucky, we have at least 32-bit signed integers in our data model, so we could avoid dealing with endianness by using shift or multiply to extract bytes out of it. We could probably do that with Retro Forth with some rewriting.

   • Paolo Redaelli (paoloredaelli@mastodon.uno)'s status on Saturday, 11-Jun-2022 17:03:32 UTC Paolo Redaelli

     in reply to

     @natecull actually only bytes matters but you must know the endianness of the encoding. Which bit is the most significant, the first or the last?

   • Santa Claes 🇸🇪🇭🇰🎅 (clacke@libranet.de)'s status on Saturday, 11-Jun-2022 17:30:17 UTC Santa Claes 🇸🇪🇭🇰🎅

     in reply to
     @natecull From your three examples, by "fundamental abstractions" and "means of data retrieval" do you mean literal expressions/encodings of data types?
     
     So "dictionary" wouldn't be relevant for the list because it doesn't say how it looks in RAM, but "dictionary stored as the RFC 8949 'map' type" might make it to the list, if expected to be around by 2122, because it does say how it looks in RAM?
   • Nate Cull (natecull@mastodon.social)'s status on Sunday, 12-Jun-2022 05:18:19 UTC Nate Cull

     in reply to

     • Santa Claes 🇸🇪🇭🇰🎅

     @clacke

     My concern is less about "how it looks in RAM" and more "does it have very precisely defined semantics".

     "Dictionary/map/hash" does not have precisely defined semantics in the same way that "integer" or "array" do, in my experience. There are a vast number of different - and all subtly incompatible - dictionary/map/hash definitions.

     Good luck transferring data that's a Lua object into a Javascript object, for example. You could create a coding scheme, but it's not 1-for-1 compatible.

     Santa Claes 🇸🇪🇭🇰🎅 likes this.