Re: Hardcoded insert string for the event messages



Rick C. Hodgin <foxmuldrster@xxxxxxxxxxxxxxxxxxxxx> wrote in part:
On Friday, February 4, 2011 10:47:22 PM UTC-6, Robert Redelmeier wrote:
Yes, today. But when all this was being standardized, the 1960s,
IBM was the main pusher for 32bit = 4 * 8. Others used 24 or 36
bits, 6 bit "bytes" and octal transcription rather than hexdecimal.
"byte" is not unambigiously 8bit, so standards say "octets".

x86 is ubiquitous. Intel is ubiquitous. ARM is ubiquitous.
And all data interchanged on communication pathways comes in
the form of 8-bit bytes of data, converted to whatever other form.

There are always other forms. But what's most prevalent?


Most prevalent does not imply "should be used". It is a
tyranny of the majority to assume or impose CP437 upon users
who have very legitimate reasons for using something else.
Yes, CP437 has cute line-drawing, math and other symbols but is
missing characters for some European languages. If you need
brevity, stick with 7bit ASCII.

ASCII is locked at 7bit by history much the same way the
x86 instruction set encoding looks so funny. Not its'
non-orthogonality, I'm not a compiler but an asm-coder and
orthogonality is almost useless to me. Have a look at an
opcode map. One byte opcodes wasted on seldom-used instructions.
Could have forced them into two bytes and used the saved opcode
bits for extra registers or extra addressing modes. But no,
that was _NOT_ possible since 8080 compatibility [portability]
was a major constraint. Also explains some of the flag quirks.

Standards have big benefits otherwise they would not be used.
Lock-in is their cost. Do not think you can ignore the cost.


-- Robert

.