Conversation

Notices

1. Embed this notice
   Philipp :emacs: :nixos: (pmidden@fosstodon.org)'s status on Friday, 08-Nov-2024 18:35:12 JST Philipp :emacs: :nixos:

   #haskell's time library is extremely future and past-proof using unbound integers: "type Year = Integer". On second thought, maybe interesting for astronomists and geologists?

   • Embed this notice
     Alfred M. Szmidt (amszmidt@mastodon.social)'s status on Friday, 08-Nov-2024 18:35:10 JST Alfred M. Szmidt

     in reply to

     • Weekend Editor

     @weekend_editor (only when it wasn't ... TIME and such protocols it would be a fixnum, and inlow level code it would also need to be a fixnum so it could be unboxed/boxed :-) )

     @pmidden

   • Embed this notice
     Weekend Editor (weekend_editor@mathstodon.xyz)'s status on Friday, 08-Nov-2024 18:35:12 JST Weekend Editor

     in reply to

     @pmidden

     On Lisp Machines, and hence now in Common Lisp, time was typically measured in "universal time", i.e., time since 1900 in seconds.

     It is an unbounded integer (bignum) for the same reason.

   • Embed this notice
     Alfred M. Szmidt (amszmidt@mastodon.social)'s status on Friday, 08-Nov-2024 23:57:38 JST Alfred M. Szmidt

     in reply to

     • Weekend Editor

     @weekend_editor @pmidden I can't comment on Symbolics stuff, it used a internal clock chip or something to get the time instead of (by default) asking the network.

     The CADR had no real RTC ... it would ask the network for the time (32-bit) and then be happy, using a 60Hz tick from the electrical network to keep track (I think).

     I had to fix a bunch of places in the System code to manage 2000 ...

   • Embed this notice
     Weekend Editor (weekend_editor@mathstodon.xyz)'s status on Friday, 08-Nov-2024 23:57:39 JST Weekend Editor

     in reply to

     • Alfred M. Szmidt

     @amszmidt @pmidden

     Network protocols are another matter, since they have to be understandable by a foreign machine of unknown architecture.

     But as far as I recall (admittedly it's been many years), internal representations of time on Symbolics machines were all bignums.

     Boxing/unboxing wasn't such an issue since the hardware had type tags and would handle a lot in microcode.

     But, as I said, it's been 40+ years since I had to know this in any detail.

   • Embed this notice
     Alfred M. Szmidt (amszmidt@mastodon.social)'s status on Friday, 08-Nov-2024 23:58:53 JST Alfred M. Szmidt

     in reply to

     • Weekend Editor

     @weekend_editor @pmidden Other silly places is where you have the QFASL (or QBIN ..) where the file property list has a fixnum with the creation date, that isn't stored in bignum for obvious reasons.

   • Embed this notice
     Alfred M. Szmidt (amszmidt@mastodon.social)'s status on Saturday, 09-Nov-2024 00:07:10 JST Alfred M. Szmidt

     in reply to

     • Weekend Editor

     @weekend_editor @pmidden They still, I think, had the :creation-date encoded in them as a fixnum -- though maybe it could be a digit that was a big number and got handled later ...

   • Embed this notice
     Weekend Editor (weekend_editor@mathstodon.xyz)'s status on Saturday, 09-Nov-2024 00:07:11 JST Weekend Editor

     in reply to

     • Alfred M. Szmidt

     @amszmidt @pmidden

     The .bin file for the Symbolics L machines was reworked a couple times; this may have been changed with respect to the CADR/LM-2.

   • Embed this notice
     Alfred M. Szmidt (amszmidt@mastodon.social)'s status on Saturday, 09-Nov-2024 00:08:22 JST Alfred M. Szmidt

     in reply to

     • Weekend Editor

     @weekend_editor @pmidden

     "One day, Doug Dodds (I think?) came up with a "solution". (Quotes because it worked for us, with a lot of lispms, but wouldn't work for a customer site with only 1 or 2.) It was to have the lisp machines all ask each other what time it was, and then set their clocks to the average of the result. "

     NTP before it was invented 😉

   • Embed this notice
     Weekend Editor (weekend_editor@mathstodon.xyz)'s status on Saturday, 09-Nov-2024 00:08:23 JST Weekend Editor

     in reply to

     • Alfred M. Szmidt

     @amszmidt @pmidden

     Yes, we had internal clock chips on the 3600.

     But they weren't very good! There was significant forward/backward drift in seconds as the days went by. And since those machines were reliable enough that they might not be booted for months, this was a problem.

     One day, Doug Dodds (I think?) came up with a "solution". (Quotes because it worked for us, with a lot of lispms, but wouldn't work for a customer site with only 1 or 2.) It was to have the lisp machines all ask each other what time it was, and then set their clocks to the average of the result.

     Some time chips were faster, some were slower, but it averaged out *quite* nicely.

     I had a nice time explaining the Central Limit Theorem to some colleagues who wanted to know why it worked. :-)

   • Embed this notice
     Alfred M. Szmidt (amszmidt@mastodon.social)'s status on Saturday, 09-Nov-2024 00:09:37 JST Alfred M. Szmidt

     in reply to

     • Weekend Editor

     @weekend_editor .. HOSTS TEXT existed before that, and HOSTTAB as a protocol.

     @pmidden

   • Embed this notice
     Weekend Editor (weekend_editor@mathstodon.xyz)'s status on Saturday, 09-Nov-2024 00:09:38 JST Weekend Editor

     in reply to

     • Alfred M. Szmidt

     @amszmidt @pmidden

     And NAMESPACE was DNS before *that* was invented, too.