[rfc-i] Digital Preservation Considerations for the RFC Series -- draft-flanagan-rfc-preservation-00.txt is posted

Paul Kyzivat pkyzivat at alum.mit.edu
Wed Sep 10 09:33:57 PDT 2014

On 9/10/14 11:59 AM, Dale R. Worley wrote:
>> From: Paul Kyzivat <pkyzivat at alum.mit.edu>
>> Unfortunately ASCII is a problem on punched cards.
>> (The good old days when characters only required 6 bits!)
>> Was there ever a standard punched card format for ascii?
> Tsk!  It's even in Wikipedia:
>      ANSI X3.26 – 1980/R1991) Hollerith Punched Card Code

Hmm. I'd stopped using punched cards by then.
I wonder if anybody ever used cards in that format?

> Unfortunately, the document is still locked up by ANSI, but the code
> itself is documented about midway down
> http://www.quadibloc.com/comp/cardint.htm  That page shows a full
> 8-bit punch card code.  Hmmm, which means it can be used to punch
> UTF-8 representations of Unicode...
>> From: Ted Lemon <mellon at fugue.com>
>> (Seriously, card punch machines were a lot of fun for me, because
>> they were already pretty steampunk when I first used them.   But
>> storing RFCs on punch cards is probably not the way to go, sad
>> though it may be to acknowledge this.)
> Tsk!  I read in
>      http://www.ietf.org/id-info/checklist.html
>      2.1.  Formatting
>      [...]
>      1. No text beyond the 72nd column of a line. This is especially
>         important for diagrams and code, which the RFC Editor may not be
>         able to trivially reformat to fall within the margins.
>      [...]
>      5. ASCII-only, no control characters (other than CR, NL and FF).
> And where do you think the 72-column rule came from?  This is so the
> line can be centered on an 80-column display, which is a lineal
> descendent of the 80-column Hollerith card.

No. The common convention for use of 80 column punched cards was that 
columns 73-80 were used as a sequence number for the cards in the deck. 
(Like a line number.) So when you were punching Fortran cards, the 
compiler only considered columns 1-72 to be significant. This convention 
was widely followed for most uses of punched cards. So a line length of 
72 became a magic number.

I don't *know* that is why RFCs have a 72 char limit, but I think it likely.

> Indeed, the RFC series is probably the *only* modern document series
> that can reliably be punched on cards!



More information about the rfc-interest mailing list