Re: The correspondence of the command-name and its key-sequence.

[Omar Polo <op@omarpolo.com>]

Hongyi Zhao <hongyi.zhao@gmail.com> writes:

> On Sat, Jan 9, 2021 at 4:23 PM Omar Polo <op@omarpolo.com> wrote:
>[..]
>>
>> You shouldn't mix the hexadecimal and decimal representation of the
>> characters.  The ASCII characters 0x01-0x1a (inclusive), i.e. the
>> characters in the range 1-26 (decimal), are the ctrl- variant of the A-Z
>> ASCII characters (0x41-0x5a).
>
> Why is the ASCII table designed into this strange layout?

I have no clue.  I wasn't there when they designed it :)

>> I don't remember if the following is right, so please correct me if I'm
>> wrong, but the keyboards (the old ones I mean) when pressing the ctrl-
>> key along with a letter would apply a bitmask on that key.  If you look
>> at this wikipedia page[0], you'll see that the ASCII character 0x01
>> "Start of Heading" is mapped to C-a.  Now, if you look at the byte
>> representation of the 'A' character in ASCII it is
>>
>>    4    1       (decimal)
>>   0010 0001     (hexadecimal)
>>
>> and that SOH is
>>
>>    0    1       (decimal)
>>   0000 0001     (hexadecimal)
>>
>>
>> and so on, B is mapped to to STX, C to ETX, ... G to the BELL character,
>> ... I to the hard tab.
>
> The `man acsii' says the following:
>
> HT  '\t' (horizontal tab)
>
> Why you describe it as the hard tab?

A Freudian slip.  I intended to write horizontal tab :)


>> So it seems that CTRL(key) is (key & 0xDF).
>
> I still don’t quite understand how this is derived.

it clears the 3rd bit (counting from left to right).  For instance, for
I <-> tab you get

   4    9       (decimal)
 0010 1001      (binary)

which becomes

   0    9       (decimal)
 0000 1001      (binary)
   ^
   this one

As you can see, that bit get cleared.  0xDF is 1101 1111, so it should
"select" every bit but that one.

Note that this works even after the Z.  [ (that's right after the Z)
gets mapped to C-[, which is ESC.  \ gets mapped to "field separator" etc

( [ is 0011 1010, ESC is 0001 1010 )

>> (Again, correct me if I'm
>> wrong.  I recall reading something like this time ago, but I couldn't
>> find the source now)
>>
>> Now, exactly why C-g was chosen for to the keyboard-quit, is something I
>> don't know, but the fact that by default it produces a BELL character
>> (on ttys at least) may be a hint.
>>
>> [0]:https://en.wikipedia.org/wiki/C0_and_C1_control_codes#Basic_ASCII_control_codes
>
> Thanks a lot for this interesting and presumably feasible analysis.
>
> Best,