ascii dates

[Anonymous <nobody@replay.com>]

On Sat, 20 Feb 1999 08:40:25 -0800, "Stephen Gutknecht (vw)"
<VW@i405.com> had the foolishness to write:

> ...It wasn't good programming to store dates in 2
> bytes instead of 4.
> 
> ...it was GOOD PROGRAMMING to store years as ASCII or
> EBC** instead of...binary?  That back then when
> a bit was a lot, they didn't understand how to use
> bits?

Ah, another brainless young loudmouth who thinks his narrow
present view of reality empowers him to understand in
myopic retrospect any and all of history -- someone who has
NO CLUE how business data processing systems were designed
and built before his time, that many early DP computers did
LITTLE OR NO pure binary operations, that they were BUILT
to do arithmetic directly on fields of encoded decimal
digits.

> Last time I checked, you can store 256 values in a
> single byte.  You don't need...two ASCII...digits.

Last time I checked, you were clueless. There _were_ no
"bytes," asshole, and there was no such thing as "ASCII"
when this "problem" began, you insufferable juvenile
asswipe.

Trying to make a case for four-digit dates in the 1950s,
60s, even 70s, would have been like trying to make a case
for strong crypto in email, or for wisely chosen
passphrases -- completely irrelevant.

You should work on your education before mouthing off
in public.
  What was BCD?
  Where did the IBM 1620 get its multiplication table?
  How was _all_ business data processing done at one
   time? (hint: card stock and holes)

> It is poorly trained programmers and the general
> infancy of the industry that lead to this.  Their
> weren't standards for databases (such as SQL) and
> it was norm for anyone.

No, it is poorly educated people who _think_ they
are programmers and their infantile nature that lead to
comments like yours. There _were_ no standards for
databases because there _were_ no fucking "databases,"
shitbrain!

> I say that it is more to do with saving $ and
> lack of understanding the cost of sitting still
> vs. the cost of re-working a system.

_I'd_ say that you haven't the foggiest idea, one way
or another. You weren't there, but you are full of
unilaterally-developed opinions, a sure sign of Motor
Mouth Syndrome.

You were replying to dana, who wrote:

> Ehem.  Nice try, but it was caused by 2k computers,
> 10M harddrives (or no harddrive), and...

Dana, you're in better shape than Sunder or vw, but even
your examples are based on fairly recent technology.

> I would bet that over...30 odd years the savings from...
> 2 digit years would...pay for this "crisis".

Nice try, but no such cost justification was ever done.
If you live as a guest in your brother's studio apartment,
you don't buy an 18-foot sectional. If you drive a VW Bug,
you don't invite six friends to go for a drive with you.
If you were programming during the first 40 or so years of
the data processing era, you never gave any thought to 
tying up two extra digits.

There _were_ no calculations; there _was_ no discussion.

> Some...Company...has...data...date in 4 fields...
> By only using 2 digit dates you are able save 64 bits

Digits did not occupy 8 bits when this started. IBM's
BCD used six bits per character. Later, in 8-bit byte
machines, numerics were often packed down to 4
bits per digit. The hardware was able to do arithmetic
with packed decimal values. Binary was difficult or
impossible, always a pain in the ass.

> Now in the late 80's I bought a 40M scsi harddrive
> and it cost me $900.

When this "problem" started, there _were_ no "hard
drives." Think tape: 7-track, 200 bpi tape. Think
dozens of $50,000 tape drives taller than you.
Think doing a sort in 14 hours with input & output
tapes and six or eight drives for "work" tapes. Think
hours of mounting and dismounting tapes. Think acres of
tape racks. Picture Sunder and vs shooting themselves
in the head in depression if they had to deal with that.
Think how important Y2K would have been in that context:
zip.

"Hard drives?" How about the 80MB and 300MB removable
platter drives of the late 1970s and into the 1980s?
How about $20,000 for the smaller one and $30,000 for
the larger one? For perspective, those numbers would
be about $44,000 and $66,000 in 1998 dollars.

Early rotating magnetic storage was much smaller, of
course, not yet available at all in early DP. Drum
memories, predating disk memories, had sizes
measured in Kwords, not Mbytes. The word size usually
matched the wordsize of the CPU.

You mention:

Paper tape

  Widely used in teletype machines. 5-level Baudot.

  8-level tape came later, in the late 1960s, and
  came to be used in TWX machines and computers.

Punched cards

  80 columns, 12 bits per column. Used for quite
  a time as the _primary_ data medium. Updates
  were done by reading an input deck, modifying
  the data, and punching a new output deck. The
  input deck would be thrown away after the
  output was seen to be good.

8" floppys

  Too recent.

<100M harddrives

  Same here. Early rotating storage was in K, not
  M. If you had proposed 4-digit years you would have
  lost your job.

Nine and sixteen track tape

  16-track? 18-track came _much_ later, as a SCSI
  device. Even 9-track is "new" stuff. The first
  widely-deployed standard for tape was 7-track, with
  an XOR parity checksum. Think many, many, erroneous
  reads due to self-correcting multiple bit errors.
  Lots of fun.

Floppy jukeboxs

  New stuff.

The first computers used in launch guidance at Cape
Canaveral had memory bits the size of your fist and
were hand-programmed: for a 0, pass the wire through
the doughnut in this direction -- for a 1, pass the
wire through in that direction. Total memory something
like 176 bits.

Some of the really slick stuff of the 1950s used
recirculating drum tracks... a read head passing
bits from the drum through to a write head on the
same track, the data in constant recirculation.
Track size a function of physical placement of read
and write heads around the drum. Transfer to and from
other circuitry done by intervening in the bit stream.
Registers as short tracks recirculating once for
every word in the long tracks, the serial bit streams
gated together for arithmetic and logicial operations,
the result gated to a write head.

19 tracks, 108 words per track, 28 or so bits per word
-- the total of main memory. Think of writing numeric
code by hand, of calculating the execution time,
figuring the locations on the track of the instruction
and its operands. Each instruction actually pointed to
the next instruction -- the only way to get reasonably
fast program execution. No assembler. No compiler.

Someone eventually wrote a compiler, no doubt
by _hand_. The machine was used for complex civil
engineering calculations. A _huge_ library of
freely exchanged scientific and engineering software
was built by user/contributors.

Oh yes... the machine was 100% vacuum tubes, with a
MTBF of, oh, weeks, no more than a few months.

The world of vw's parents, perhaps Sunder's parents,
was built and run on equipment like I describe above.
And they sit here in P-500 Land with integrated
development environments and virtually no knowledge
of _any_ of the underlying hardware, and _whine_
about how the previous couple of generations got
them here.

Sheesh!

Born Too Early To Have _REALLY_ Had Fun