Re: Top 10 most common hard skills listed on resumes...

Michael S <already5chosen@yahoo.com> writes:

On Fri, 06 Sep 2024 07:56:56 -0700
Tim Rentsch <tr.17687@z991.linuxsc.com> wrote:
>

Michael S <already5chosen@yahoo.com> writes:
>

On Fri, 6 Sep 2024 10:35:16 +0100
Bart <bc@freeuk.com> wrote:
>

On 05/09/2024 22:37, James Kuyper wrote:
>

On 9/5/24 12:54, Kaz Kylheku wrote:
>

On 2024-09-05, Waldek Hebisch <antispam@fricas.org> wrote:

>
...
>

You seem to miss the point that assigment operator is
fundamentally assymetic.

>
Both sides of an assignment can be complex expressions that
designate an object (though the right side need not).

>
So you've correctly identified the very fundamental asymmetry.

>
Sure, if you want to completely disregard all the cases where the
symmetry does exist.
>
That means that for you, there is no interesting difference (using
my example of assigning A to itself) in a language where you write
'A = A', and one where you write 'A = .A'.
>
(I'd be interested in how, in the latter language, you'd write the
equivalent of 'A = A = A' in C, since the middle term is both on
the left of '=', and on the right!)

>
The point is that in BLISS everithing that is legal on the right
side of asignment is also legal on the left side.
I don't know if the point is generally true.  In particular, if
BLISS supports floatig point, what is meaning of floating point on
the left side?

>
BLISS is word based and typeless.  On a PDP-10, doing a
>
    .pi = 0
>
where 'pi' holds a 36-bit floating-point value (and 3.14159...
presumably), that floating-point value would be used as an
address and 0 would be stored into it (assuming I remember
BLISS correctly).

>
On PDP-10 reinterpreting [18 LS bits of] floating-point as address is
natural, because addresses, integers and FP share the same register
file.
It seems to me that on S/360 or CDC-6K or PDP-11 or VAX it would be
less natural.

I don't think one thing has much of anything to do with the
other.  It seems just as unlikely to use a floating-point value,
or a portion of a floating-point value, as an address on a PDP-10
as it does on any of the other systems you mentioned.

However, natural or not, BLISS was used widely both on PDP-11 and on
VAX, which means that it worked well enough.

BLISS is, or was, closer to the hardware than C.  Also it's
harder to scale in BLISS than in C, because BLISS is typeless.
Writing code in BLISS needs more discipline than writing in C.
(Disclaimer:  my experience writing code in BLISS is very
close to epsilon, and is very much dimmed by the long passage
of time.)

So probably not what one wants to do. ;)

>
Yes, LS bits of FP as address do not sound very useful.
On the other hand, using several MS bits of FP, although typically
fewer than 18, as address is useful in calculations of many
transcendental functions.

Probably not use it as an address but rather as an index.
Perhaps something like this (please forgive the bastard
mixing of BLISS and C):

    // variable d contains a 64-bit double

    needed = .(most_digits + (.d >> 52 & 0x1FFF))

to index a table 'most_digits' by the 11-bit exponent of a
"double" floating-point value.

I don't remember enough BLISS to know how to write indexing,
but this construction should accomplish that.