#RetroComputing Meetup @cbase
Freitag, 04.04. ab 19 Uhr
in der c-base, Rungestraße 20, 10179 Berlin

Offener #VintageComputing Stammtisch

Bei einem Getränk reden wir über Themen von #Zuse bis #Amiga, egal ob #Atari, #Commodore, #Apple oder #PDP8-#PDP11 - Hauptsache alte Rechner

Everything else has FTPed over to the Minitel/PiDP8i evening stereo without any trouble apart from Floral Shoppe. A problem with the Japanese characters? I've moved a tonne of City Pop over without any issues.

Finally moved it physically with a USB drive. We might have expected this from Vektroid.

Makes no difference to how it looks, but I upgraded the microSD that the PiDP8/i runs on from 32GB to 256GB, which gives it some decent storage capacity in its main role as the evening stereo system.

FTPing music from my main desk PC seems to work pretty well. It's almost like a feasible solution.
#minitel #pdp8 #pidp8i #retrocomputing #MusicForLeisure

#RetroComputing Meetup @cbase
Freitag, 07.03. ab 19 Uhr
in der c-base, Rungestraße 20, 10179 Berlin

Offener #VintageComputing Stammtisch

Bei einem Getränk reden wir über Themen von #Zuse bis #Amiga, egal ob #Atari, #Commodore, #Apple oder #PDP8-#PDP11 - Hauptsache alte Rechner

Okay, the Minitel 1 is connecting to the PiDP8-i as a terminal. That wasn't something I was expecting to figure out before lunchtime.
#pdp8 #pidp8i #raspberrypi #minitel #retrocomputing

Some TSS8 #pdp8 sources have been found!

https://forum.vcfed.org/index.php?threads/dectapes-and-linctapes-a-pile-of.1251772/

I have a slightly absurd #RetroComputing project for the #PDP8 that I find kind of fun because it forces me to find new and novel data structures and algorithms for things. You can't Just Use A Hash Table on a system with 12-bit values, 8k of core, and no multiply. It's hilarious finding algorithms that say "this structure is good for space-constrained systems, because the main table only takes up 64kB of RAM" and realising how over spec that is.

So I was digging through disco-era algorithms and data representations, mostly. Lots of neat stuff, some overlooked gems (I have a cool variable-length four-bit text representation that I found via wikibooks). But text processing is still pretty memory-hungry, with some huge tree structures necessary for the heavy lifting, so I'm mostly building abbreviation tables and such via python scripts.

But then I found LCP arrays, which I had missed for some reason. I mean, not "some reason" but that they were actually published after my university algorithms books were written! They're not like SUPER new, but most of the work on them was done by bioinformatics nerds in the early 2000s, so I didn't really notice them until now.

They're super neat! You just create a suffix array by sorting indices into your string lexicographically (from index to the end), then you go down that in order and compare each substring to the one before it and record the length of the largest common prefix in that. Bingo! Two arrays the size of your working string, and you can do all sorts of queries on them in linear time to find valuable repeated sequences for compression tools like LZ77 or an abbreviation tool.

It really *feels* like a 1970s algorithm, but back then they were all doing this in like O(N*log(N)) via dynamic programming techniques. I may still bubble sort just to save code size...

#RetroComputing Meetup @cbase
Freitag, 07.02. ab 19 Uhr
in der c-base, Rungestraße 20, 10179 Berlin

Offener #VintageComputing Stammtisch

Bei einem Getränk reden wir über Themen von #Zuse bis #Amiga, egal ob #Atari, #Commodore, #Apple oder #PDP8-#PDP11 - Hauptsache alte Rechner

Ein PDP-8 Emulator auf dem #Z80 für #SymbOS und das auf dem #Amstrad #CPC. Sehr cool ! #retro #pdp8 https://youtu.be/0FEoEv5Kpes?si=oLAgXEjYUJ-deCbX

#RetroComputing Meetup @cbase
Freitag, 03.01. ab 19 Uhr
in der c-base, Rungestraße 20, 10179 Berlin

Offener #VintageComputing Stammtisch

Bei einem Getränk reden wir über Themen von #Zuse bis #Amiga, egal ob #Atari, #Commodore, #Apple oder #PDP8-#PDP11 - Hauptsache alte Rechner

#RetroComputing Meetup @cbase
Freitag, 06.12. ab 19 Uhr
in der c-base, Rungestraße 20, 10179 Berlin

Offener #VintageComputing Stammtisch

Bei einem Getränk reden wir über Themen von #Zuse bis #Amiga, egal ob #Atari, #Commodore, #Apple oder #PDP8-#PDP11 - Hauptsache alte Rechner

Dear #RetroComputing #PDP8 enthusiasts, do any of you have a neglected #PiDP8 that you no longer use? Was it a fun lockdown toy that you no longer interact with? Are you within postal distance of #London? If so, get in touch: I'm working on a project in PAL assembler and would love to have the front switches and lights, but not enough to order and assemble the thing.

I can replace the #SBC inside, if you already scavenged it for something else! I just wonder if anyone has one taking up space that they'd rather use for something else these days.

Okay, I've got a page (128 words) of #PDP8 code that will generate a truncated Soundex hash of any ASCII or 6-bit TEXT string.

Since I have only 12 bits to store it, and the consonant classifiers are from 1-6, that takes 3 bits per digit. I'm using 5 bits to store the initial, so I get two digits plus a final bit just to distinguish if there was a third digit or not.

I coded this up in awk first to test it, with a giant `switch` statement (only in gawk, alas) mapping letters to digits. But in PAL assembler for the PDP-8, I decided instead to make a table like so:

```
ALPHBT, 2170; /CBA@
2173; /GFED
2277; /KJIH
7554; /ONML
2621; /SRQP
7173; /WVUT
0272; /[ZYX
0000; /_^]\
```

Basically the characters in this section of the character set are there in reverse order. The top three bits of the 5-bit char values indicate which word of this table to load in, and the last two indicate how many times to shift right by 3 before masking off the 3 least-significant bits. So the letter D is `04`, which means we get word `1` and shift it `0` times before pulling out the octal digit `3` (which is correct!)

Most of the code is tests for various bit patterns to see if we need to keep shifting or if we care (is it even a letter? Is it a vowel? Is it the same digit we saw last time? `H` and `W` are special cases...)

EDIT: I am not looking to set up TreeSitter on neovim. I have that working fine, but only for bog standard 2024 syntax like awk or C or Makefiles. I want a *new* syntax definition for an old assembler that targets 12-bit minicomputers from the 1960s.

Hey, are there any #NeoVim nerds here who can help folks with slightly older versions? I have a goal of "don't make my vi installation a full-time job" and I'm doing a lot of #PDP8 assembly in PAL. I can supply a #Kate syntax file that does well enough (I use it for #pandoc), but I need a #TreeSitter syntax wotsit that I can just Install without having to upgrade NeoVim or TreeSitter past what's in Debian (this is non-negotiable: anything else has always become a full-time job for me).

All the TreeSitter tutorials say to use an init command that doesn't exist in my copy, and I lost patience right then. The architecture makes it seem like this is only a pain for setting up a new project, but I don't have the attention to devote to this. I'd also like it if PAL8 syntax made it into whatever mainstream syntax repositories TS grabs from.

TFW think you have an output bug running your #PDP8 PAL code under #OpenSIMH, and then it turns out you just did a `HLT` right after printing the final character, and the emulator had stopped EVERYTHING before the pretend teletype had finished printing.

```
DONE, TSF
JMP .-1 /WAIT FOR TELETYPE READY
HLT
```

FIIIINE.

I've been tinkering with ways to squeeze slightly more modern #InteractiveFiction onto a #PDP8 over the past year or two. I was playing with the dictionary from the #Hibernated 1 Director's Cut, which has about 650 entries, and managed to come up with a system that distinguishes all but 55 of those uniquely, and those 55 are lumped into 16 buckets.

Not a bad collision rate, especially given the side-effects of my scheme are less surprising to human players than some other approaches!

The only drawback is that it can't distinguish the diagonal directions well, but with careful command choices and some workarounds in special cases we can avoid this and still have a fairly plush-feeling vocabulary. Now I just need to work on the rest of it to see if it was worth the brainwave!

Found this in an old pile of papers. I didn't realise I still had this!