Search
Saturday, 3rd of April 2021, 21:37:58 UTC
21:38:33
jcowan
I cut my teeth on OS/8, which was 8+2 with no hierarchy
21:38:42
pjb
Well, it wasn't really a restriction, it was an optimization. Implemented in LISP 1. You can see it in the sources of LISP 1.5.
21:39:30
pjb
Basically, 36-bit hardware, 6-bit characters. Strings were represented using lists of words containing packed 6-character.
21:39:43
aap
ITS filenames are literally 6 + 6
21:39:46
pjb
So if you used more than 6-character symbols, they used more than one word.
21:39:56
aap
usually filename + extension
21:40:18
aap
so it would have been extremely unusual to call something by a name longer than 6 chars
21:40:39
pjb
unix too was developped on such a system (18-bit words), hence creat and similar function names.
21:41:01
aap
pdp-7 unix used 2 chars per word
21:43:10
pjb
That said, indeed, one could wish more people used more unique names. I liked IBM naming scheme. Like IEBFR14; there's no ambiguity.
21:43:32
pjb
On the other hand, try to search for "true"…
21:44:01
pjb
You need to add eg. true man page for some semblance of uniqueness and specificity.
21:44:26
aap
does an empty file need a manpage? :)
21:44:40
aap
i guess it's no longer an empty file these days...
21:45:08
pjb
aap: the story of true matches the story of iebfr14. You'd be surprised by the commit list.
21:45:50
pjb
https://github.com/coreutils/coreutils/commits/master/src/true.c
21:47:28
aap
true used to be an empty file
22:21:46
pjb
aap: an empty file was clearly a bug for true(1)…
22:22:29
phoe
why are we discussing true(1) again
22:23:45
pjb
phoe: naming schemes iebfr14 vs true
23:29:10
jcowan
aap: they should have used UTF-9
23:29:40
aap
well, ken did eventually invent utf-8 at least :)
23:30:00
jcowan
https://tools.ietf.org/html/rfc4042
23:30:37
aap
these were more of a joke, right?
23:31:42
jcowan
CHeck out the publication date. BUt Crispin really was a PDP-10 programmer
23:32:12
aap
oh i didn't even realize it was from him
23:32:23
jcowan
UTF-18 should be replaced by UTF-18+, which would represent planes 0,1,2,3 instead of 0,1,F,10
23:34:38
jcowan
Gotta love this: "We are now in the process of implementing support for nonet-based text files and automated transformation between septet, octet, and nonet textual data."
23:34:59
aap
i'm not even sure whether this is a joke
23:35:10
aap
they're still using pdp-10s at XKL
23:35:10
aeth
UTF-18+ sounds like a media rating scheme in some random country.
23:35:22
aap
although i hear they're using a custom OS these days
23:36:56
jcowan
Compuserve ran a modified TOPS-10 for a very long time, which is why Compuserve emails were 7xxxx,xxxx@compuserve.co,m
23:38:57
jcowan
I also devised UTF-6, an almost upward compatible extension of Sixbit, but I didnt write an RFC, so I've forgotten the details
23:39:27
aeth
What Common Lisp do people here use for PDP-10s?
23:39:44
aap
no common lisps unfortunately. but maclisp!
23:40:47
aap
you can all come to #pdp-10 if you want :)
23:42:02
aeth
But why would I run a PDP VM when I could just run a Windows 98 VM?
23:43:31
aeth
I wonder if SBCL 0.6.8 runs on Windows 98
23:44:08
aap
actually implementing a common lisp for PDP-10 would be interesting
23:44:15
aeth
(hmm, 0.6.8 is the oldest tagged version... SBCL news on the website goes back to 0.6.0, though)
23:45:37
jcowan
The general idea is to preempt 077, Del in ECMA-1, to show that a non-Sixbit character has been encoded in the general style of UTF-8
23:46:00
jcowan
https://www.ecma-international.org/wp-content/uploads/ECMA-1_1st_edition_march_1963.pdf
23:46:39
jcowan
I did use to assign 16-bit magic numbers for binary data using RADIX50, though
23:46:59
moon-child
jcowan: sounds like it loses the big advantage of utf8, which is that you can always identify continuation bytes (sextets?)
23:47:22
moon-child
(though there may not be very much you can do about that ...)
23:47:22
jcowan
Yes. Perhaps it's more like SRFI-7
23:49:43
jcowan
Ah, no, what it preempts is SI and SO, so because there are always both you can identify sequences.
23:51:20
jcowan
buut it is less bounded than UTF-8
0:32:05
jcowan
It should be possible to run CLISP using one of the several C compilers avaiilable for TOPS-10/20.
0:32:25
jcowan
JIT is of coourse another story altogether
0:32:35
jcowan
(or native-code AOT)
1:10:02
Aurora_iz_kosmos
** NICK Aurora_v_kosmose
2:25:25
russell--
** NICK Guest73633
3:04:34
beach
Good morning everyone!
3:06:21
beach
jcowan: The "estimate" part is exactly that, i.e., branching.
8:58:16
Guest73633
** NICK russell--
Sunday, 4th of April 2021, 9:37:58 UTC