13 Feb
2021
13 Feb
'21
3 p.m.
Thank you for banner! I used the data, abliet modified, 40 years ago
in 1981, for a banner program as well, on an IBM 1130 (manufactured 1972)
so it could print on an 1132 line printer. The floor would vibrate
when it printed those banners. I used "X" as the printed char as the
1132 did not have the # char. But those banners looked great!
I wrote it in FORTRAN IV. On punched cards. I did this because
from 1980-1982 I only had access to UNIX on Monday evenings from
7PM-9PM, using a DEC LA120 terminal, it was slow and never had
enough ink on the ribbon.
I had only 8K of core memory with only EBCIDIC uppercase so there
were lots of compromises and cleverness needed -
- read in a 16-bit integer as a packed two 8-bit numbers
- limit the banner output to only A-Za-z0-9 !?#(a)'*+,-.=
- unpack the char data into buffer and then process it.
- fix the "U" charater data
- find the run-lenght ecnodings that could be consoldated to save space
(seeing those made me think it had to have been generated data)
The program still survives here - http://ibm1130.cuzuco.com/
(with sample output runs)
Also since I had to type all those numbers onto punch cards
with a 029 keypunch, to speed things up I coded my own free-form
atoi() equivalent in FORTRAN, reading cards, then packed two numbers into
a integer, then punch out those numbers along with card ID numbers in columns
73-80 on the 1442. This was many weeks of keypunching, checking,
fixing and re-keypunching.
That code is here http://ibm1130.cuzuco.com/ipack.html
When done the deck was around 8" or so. It took well over a
minute to read in the data cards, after complition.
Again thanks! Many hundreds of banners for many people were printed
by this, around 2 to 3 a week, until July 1982, when that IBM
was replaced by a Prime system. I still have many found memeories of
that 1130.
-Brian
Mary Ann Horton (mah at mhorton.net) wrote:
We had vtroff at Berkeley around 1980, on the big
Versatec wet plotter,
4 pages wide. We got really good at cutting up the pages on the output.
It used the Hershey font. It was horrible. Mangled somehow, lots of
parts of glyphs missing. I called it the "Horse Shit" font.
I took it as my mission to clean it up. I wrote "fed" to edit it, dot by
dot, on the graphical HP 2648 terminal at Berkeley. I got all the fonts
reasonably cleaned up, but it was laborious.
I still hated Hershey. It was my dream to get real C/A/T output at the
largest 36 point size, and scan it in to create a decent set of Times
fonts. I finally got the C/A/T output years later at Bell Labs, but
there were no scanners available to me at the time. Then True Type came
along and it was moot.
I did stumble onto one nice rendition of Times Roman in one point size,
from Stanford, I think. I used it to write banner(6).