[ih] RAND Unix Port code

Noel Chiappa jnc at mercury.lcs.mit.edu
Mon Feb 13 13:48:38 PST 2017

    > Has anybody figured out the "right" way to do this to maximize usability
    > (in case someone really wants to OCR or whatever)? 600DPI TIFs?
    > One-file-per-page?

What we generally do for text (engineering drawings are slightly different) on
BitSavers (what, you don't know about BitSavers? ;-) is scanned at 300 dpi in
Black-and-White and then saved as TIFFs with CCITT Fax 4 compression. (A nice
image tool called IrfanView is good for producing those, if your scanning
program doesn't produce these natively.) That produces pretty small per-page
files, which we then combine into a single PDF.

If anyone really wants to OCR this, they might want to start with Jim's
TIU TCP code (as I mentioned, the MIT machine had a copy of the source).

    > Jim's TCP was written to run on top of MOS (IIRC Micro Operating
    > System).

There were allegations that it stood for 'Mathis' Operating System'... :-)

    > The other major player in the MOS family tree was the Port Expander.

Yeah, we tried to get that running at MIT. (We had the TIU code but I don't
think ever used it.) MIT's two IMP's were maxed out, port-wise. Alas, the
people who had done the hardware interfaces on the ITS machine (we were
working with MIT-DM) had taken advantage of the fact that on the IMP side of
the DH connection, there were opto-isolators, and they'd done a short-cut with
grounding one side of the differential, or something like that. The IMP
interface cards we got from SRI (they worked with DRV11 parallel interface
cards, and did level conversion, etc) couldn't emulate _that_. So that's why
it took so long for MIT (other than Multics) to show up on the Internet -
DARPA had to buy us another IMP (one of the first C/30's), first.


More information about the Internet-history mailing list