Re: The joy of FORTRAN

On 2/28/25 9:19 AM, Scott Lurndal wrote:

c186282 <c186282@nnada.net> writes:

On 2/26/25 7:22 PM, Rich Alderson wrote:

Lawrence D'Oliveiro <ldo@nz.invalid> writes:
>

On Tue, 25 Feb 2025 15:47:15 -0700, Peter Flass wrote:

>

The idea of defining different-sized bytes is a real plus.

>

What they meant by "bytes" was really just "bitfields".

>
Look up the original definition of "byte" in the signal processing literature,
and you'll find that "arbitrary bitfield" is the original meaning, dude.
>
The restriction of "byte" to "bitfield of a particular size for the manufacturer's
architecture, especially 8 bits" is the odd choice.

>
   8 bits kinda emerged with microprocessors.

 Surely the IBM 360 was responsible for the popularity
of 8-bit bytes - and that drove the adoption by other
computer manufacturers if only to support common I/O
peripherals.

   IBM was, to a degree still is, a big and influential
   player. Lots of 360/370 systems were sold - and you
   saw 'em at big govt / NASA installations, which
   amounted to 'prestige' and a sort of a 'standard'.
   They were good boxes.
   Came across a 360 (model 20 ? the little one) in
   use by a parts distributor about 20 years ago.
   Nowadays they're always emulated, like with the
   Hercules system or better since I doubt you'll
   get parts/service for yer ancient 360/370 now.
   Bytes were bound to migrate towards some even power
   of two. Six or seven bits ... nah. 8 was kind of
   the useful minimum. 4 bits was for microcontrollers
   and some BCD calx. 12 and 18 bits were seen in
   various minis/mainframes for quite awhile.
   Hexadecimal is handy and fairly comprehensible.
   Better than octal IMHO. 8/16/32/64 works out very
   well with hex.
   I also think the die-fabricating tech when the
   first microprocessors came out made 8-bit the
   widest thing they could support easily on the
   typical amount of silicon real-estate.