It is amazing how knowledge of basic CS is absent in most
programmers, both professional and amateur alike. While
"vibe coding" is all the rage, a silent disaster is also
creeping.
What GNU/Linux requires urgently is a Decimal Floating
Point library that implements IEEE754-2008, but mention this
to most programmers and one will receive only blank stares
and gaping mouths.
The only GNU/Linux project for IEEE754-2008 is closing
its doors:
<https://github.com/libdfp/libdfp/pull/238>
For C++, Boost.Decimal.
By the way, are you AI?
Leroy H <[email protected]>wrote:
On Fri, 16 Oct 2026 11:44:54 -0400, Chris Ahlstrom wrote:
For C++, Boost.Decimal.
Some people don't use C++. I would hope that most would not.
It is recommended, and even required in some countries, that all
financial software be written using decimal FP.
But once libdfp drops out then GNU/Linux C programmers will be
stuck.
1/pi= 0.318310
q
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
You can just do decimal in C
Use type double if you are worried about microcents.
I have written financial software in C.
What is the problem?
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
I have written financial software in C.
What is the problem?
That depends on the scale.
For most situations, the currency amounts are enough that the
C double type, with 16 decimal digits of precision, can handle
the simple processing without error.
(C also has a 128-bit long double type but ordinary processors
have no hardware for this type. On GNU/Linux the long double
is usually defined as 80-bits.)
But for large currency amounts and more complex processing,
16 decimal digits may not be enough. That's why decimal floating
point is recommended or even required by law.
What is needed is HARDWARE decimal FP. IBM has always had decimal
FP hardware in its mainframes
Now that 64-bit ints are available, you could use them
in many situations.
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
But the original 360 had packed decimal, which can work with up to
31 decimal digits. I wrote a lot of business software in those days;
I can count the number of times I've used floating point on the
fingers of one hand. For basic accounting, integer pennies are
enough. Add a few guard digits if you must. To quote the old
commercial, it "takes the worry out of being close".
On Fri, 16 Oct 2026 18:10:06 GMT, Charlie Gibbs wrote:
But the original 360 had packed decimal, which can work with up to
31 decimal digits. I wrote a lot of business software in those days;
I can count the number of times I've used floating point on the
fingers of one hand. For basic accounting, integer pennies are
enough. Add a few guard digits if you must. To quote the old
commercial, it "takes the worry out of being close".
Did you ever do calculations involving financial derivatives?
Even compound interest requires calculations involving logarithms and exponentials.
Leroy H wrote this screed in ALL-CAPS (fixed):
It is amazing how knowledge of basic CS is absent in most
programmers, both professional and amateur alike. While
"vibe coding" is all the rage, a silent disaster is also
creeping.
What GNU/Linux requires urgently is a Decimal Floating
Point library that implements IEEE754-2008, but mention this
to most programmers and one will receive only blank stares
and gaping mouths.
The only GNU/Linux project for IEEE754-2008 is closing
its doors:
<https://github.com/libdfp/libdfp/pull/238>
For C++, Boost.Decimal.
The Free42 calculator provides either binary or a decimal
implementation.
By the way, are you AI?
On Fri, 16 Oct 2026 11:44:54 -0400, Chris Ahlstrom wrote:
For C++, Boost.Decimal.
Some people don't use C++. I would hope that most would not.
It is recommended, and even required in some countries, that all
financial software be written using decimal FP.
But once libdfp drops out then GNU/Linux C programmers will be
stuck.
By the way, are you AI?
Human beings have always been AI in that they can only parrot
news reports and their old school textbooks. This has been true
since pre-historic times.
Occasionally, however, some misaligned freak will rise above--
it all and then we have a burst of progress.
Leroy H <[email protected]>wrote:
On Fri, 16 Oct 2026 11:44:54 -0400, Chris Ahlstrom wrote:
For C++, Boost.Decimal.
Some people don't use C++. I would hope that most would not.
It is recommended, and even required in some countries, that all
financial software be written using decimal FP.
But once libdfp drops out then GNU/Linux C programmers will be
stuck.
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
Operator overloading is a very handy feature. Yes, C++ is overly
complicated. Java left out that feature of C++, and it, too, still
manages to be excessively complicated.
Python, on the other hand, includes operator overloading, yet is still
only a small fraction of the complexity of either Java or C++.
IMHO one of the biggest programming mistakes is using floating point
when you can do it faster, more accurately, and more easily with
integers.
Yes he is. He is Actually Ignorant.
On Fri, 16 Oct 2026 22:19:37 GMT, Charlie Gibbs wrote:
IMHO one of the biggest programming mistakes is using floating point
when you can do it faster, more accurately, and more easily with
integers.
Some languages offer direct support for “fixed-point” formats,
basically integers with a fixed scale factor, so you can do
calculations that keep a fractional part.
From what I gather, they tend to be difficult to use properly. The
problem seems to be keeping calculations within the permissible
dynamic range, to avoid overflow and underflow.
This is why floating-point tends to be far more popular.
On Fri, 16 Oct 2026 23:19:38 +0000, Nick Charles wrote:
Yes he is. He is Actually Ignorant.
Still think Apple’s phone and tablet products are running “Unix”, are you?
Lawrence =?iso-8859-13?q?D=FFOliveiro?= <[email protected]d>wrote:
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
Operator overloading is a very handy feature. Yes, C++ is overly
complicated. Java left out that feature of C++, and it, too, still
manages to be excessively complicated.
Python, on the other hand, includes operator overloading, yet is still
only a small fraction of the complexity of either Java or C++. Such a
feature is useful, for example, in letting you express operations on >mathematical objects (like vectors and matrices) using something close
to the original mathematical notation.
I never got much into Python
Never needed it for the things I do.
Also it seems to be changing all the time?
But I come from a hardware background, in the sixties there were no 'computers' yet they did a manned Moon landing and return.
IMHO one of the biggest programming mistakes is using floating
point when you can do it faster, more accurately, and more
easily with integers. That's not to say that there aren't
places where floating point is necessary - but for some of
us, that's a small portion of what we have to do. I don't
need floating point to express IP addresses, for instance.
Leroy H wrote this screed in ALL-CAPS (fixed):
Human beings have always been AI in that they can only parrot
news reports and their old school textbooks. This has been true
since pre-historic times.
They had news reports and textbooks in prehistory?
Leroy H wrote this screed in ALL-CAPS (fixed):
[.../SNIP/...]
By the way, are you AI?
Lawrence D’Oliveiro wrote this screed in ALL-CAPS (fixed):
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
Operator overloading is a very handy feature. Yes, C++ is overly
complicated. Java left out that feature of C++, and it, too, still
manages to be excessively complicated.
C++ overly complicated? Nah.
Python, on the other hand, includes operator overloading, yet is still
only a small fraction of the complexity of either Java or C++.
Prove it.
Le 16-10-2026, Chris Ahlstrom <[email protected]> a écrit :
Leroy H wrote this screed in ALL-CAPS (fixed):
[.../SNIP/...]
By the way, are you AI?
No, he's too limited for that. A modern AI would be far better in its answers.
On 17/10/2026 00:35, Chris Ahlstrom wrote:
Leroy H wrote this screed in ALL-CAPS (fixed):
Yes. And they were carried by bardsHuman beings have always been AI in that they can only parrot
news reports and their old school textbooks. This has been true
since pre-historic times.
They had news reports and textbooks in prehistory?
As songs and sagas and myths
The Bible was a written down record of some of them
Lights in the sky, great floods, cities gone in a flash. Tsunamis
causing the sea to retreat. Evidence exists for all of these. Just not
the god Yahweh. HE stared out as a minor war god. Somewhere along the
line monotheism looked to be advisable, and he got the job
The sling is well known to be a devastating range weapon when faced with
an opponent whose armour is not 100%, For example. A fact worth passing along in a neat little morality tale
The Natural Philosopher wrote this screed in ALL-CAPS (fixed):
On 17/10/2026 00:35, Chris Ahlstrom wrote:
Leroy H wrote this screed in ALL-CAPS (fixed):Yes. And they were carried by bards
Human beings have always been AI in that they can only parrot
news reports and their old school textbooks. This has been true
since pre-historic times.
They had news reports and textbooks in prehistory?
I thought it was promulgated by voice and memory....
As songs and sagas and myths
The Bible was a written down record of some of them
That's not prehistory, is it?
Lights in the sky, great floods, cities gone in a flash. Tsunamis
causing the sea to retreat. Evidence exists for all of these. Just not
the god Yahweh. HE stared out as a minor war god. Somewhere along the
line monotheism looked to be advisable, and he got the job
The sling is well known to be a devastating range weapon when faced with
an opponent whose armour is not 100%, For example. A fact worth passing
along in a neat little morality tale
I remember being amazed at how far a slingshot could sling a
marble.
(I would also shoot a BB straight up to see how close to me I
could get it to land. A couple of times me and a friend had
BB duels.)
The Natural Philosopher wrote this screed in ALL-CAPS (fixed):
On 17/10/2026 00:35, Chris Ahlstrom wrote:
Leroy H wrote this screed in ALL-CAPS (fixed):Yes.
Human beings have always been AI in that they can only parrot
news reports and their old school textbooks. This has been true
since pre-historic times.
They had news reports and textbooks in prehistory?
And they were carried by bards
I thought it was promulgated by voice and memory....
As songs and sagas and myths
The Bible was a written down record of some of them
That's not prehistory, is it?
Lawrence =?iso-8859-13?q?D=FFOliveiro?= <[email protected]d>wrote:
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
I never got much into Python
Never needed it for the things I do.
Also it seems to be changing all the time?
New features appear every year.
But I come from a hardware background, in the sixties there were no
'computers' yet they did a manned Moon landing and return.
Of course they had “computers” -- several giant IBM mainframes at
Mission Control. Also every engineer had their trusty slide rule
handy.
There were 4 computers on the Apollo spacecraft -- one to control the
Saturn rocket, one in the Command Module, and 2 (1 for backup) on the
Lunar Module.
But you are
answering The Natural Philosopher who can only close his eyes to what
goes beyond his preconceived expectations.
From what I saw, for a beginner's approach, python is easier than Java
or C++. Now when one starts to use python in the real life, the
complexity of python appears. Because that's the python way: things
works except when they stop to work.
Like, it's not a typed language but start playing with dictionaries
without declaring your variables first to see how the claims stand
before the facts.
Or you want to see what's in some variables, so you print it with print(myvariable)
When you start to have too many variables, you want to put the name of
the variable in front of it to increase its visibility, so you print("myvariable: " + myvariable)
but that doesn't work because myvariable is not a string. So you have to print("myvariable: " + str(myvariable))
But now you have the description of myvariable (or something like that:
I discovered it a long time ago, my memory isn't clear about it), not
its content, so you have to handle it differently (about that, I'm sure).
And I'm not speaking about the dependencies. Each version breaks
everything so well that now some distros don't let you install
libraries with pip if you are not in a virtual environment.
So, yes to do something simple, python is fast and easy. But if you
want to power of C++ and Java, you can't use python on a simple way.
We make extensive use of Python, apart from some performance issues (exacerbated by the not particularly speedy SoC it runs on) it’s not
really problematic.
Python is absolutely a typed language. Every object has a type and it
won’t let you use objects in a manner incompatible with their type:
>>> 0+""
Traceback (most recent call last):
File "<python-input-0>", line 1, in <module>
0+""
~^~~
TypeError: unsupported operand type(s) for +: 'int' and 'str'
10+True
<class 'bool'>type(True)
<class 'int'>type(0)
Richard Kettlewell wrote:
Python is absolutely a typed language. Every object has a type and it
won’t let you use objects in a manner incompatible with their type:
>>> 0+""
Traceback (most recent call last):
File "<python-input-0>", line 1, in <module>
0+""
~^~~
TypeError: unsupported operand type(s) for +: 'int' and 'str'
On the other hand:
❯ python3
Python 3.14.7 (main, Aug 8 2026, 20:03:42) [Clang 13.1.6 (clang-1316.0.21.2.5)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
10+True
<class 'bool'>type(True)
<class 'int'>type(0)
;-)
On 17/10/2026 15:02, Richard Kettlewell wrote:
We make extensive use of Python, apart from some performance issues
(exacerbated by the not particularly speedy SoC it runs on) it’s not
really problematic.
Wow. someone has a real use for it!
I just am too old to become proficient in another language, I am still battling JavaScript
Stéphane CARPENTIER <[email protected]> writes:
From what I saw, for a beginner's approach, python is easier than Java
or C++. Now when one starts to use python in the real life, the
complexity of python appears. Because that's the python way: things
works except when they stop to work.
Like, it's not a typed language but start playing with dictionaries
without declaring your variables first to see how the claims stand
before the facts.
Python is absolutely a typed language. Every object has a type and it
won’t let you use objects in a manner incompatible with their type:
>>> 0+""
Traceback (most recent call last):
File "<python-input-0>", line 1, in <module>
0+""
~^~~
TypeError: unsupported operand type(s) for +: 'int' and 'str'
Or you want to see what's in some variables, so you print it with
print(myvariable)
When you start to have too many variables, you want to put the name of
the variable in front of it to increase its visibility, so you
print("myvariable: " + myvariable)
but that doesn't work because myvariable is not a string. So you have to
print("myvariable: " + str(myvariable))
But now you have the description of myvariable (or something like that:
I discovered it a long time ago, my memory isn't clear about it), not
its content, so you have to handle it differently (about that, I'm sure).
It sounds like you just haven’t learned Python.
With some typed languages, you can't use a variable if you didn't
declare it's type before. With python, you don't need to declare
integers and strings but you need to declare dictionaries. Why? Don't
know.
With python, that looks easy and when I fix something, it breaks
something else for reasons I don't understand.
On 17/10/2026 18:46, Stéphane CARPENTIER wrote:
[..]
With some typed languages, you can't use a variable if you didn't
declare it's type before. With python, you don't need to declare
integers and strings but you need to declare dictionaries. Why? Don't
know.
I guess you confuse assigning a type with creating an object of a
specific type.
With python, that looks easy and when I fix something, it breaks
something else for reasons I don't understand.
I recommend to install Copilot CLI on your system and ask it to explain
why your script breaks. It's quite good at explaining is my experience.
Most code I write is in C or in asm for Microchips.
asm is extremely powerful
understand the hardware!!!
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
Most code I write is in C or in asm for Microchips.
asm is extremely powerful
understand the hardware!!!
It is good to hear someone actually standing up for
C and assembly.
These Linux groups seem to be packed only with Python/C++
jockeys. The poor dupes don't know what they're missing.
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
So, yes to do something simple, python is fast and easy. But if you want
to power of C++ and Java, you can't use python on a simple way.
On 17/10/2026 15:02, Richard Kettlewell wrote:
We make extensive use of Python, apart from some performance issues
(exacerbated by the not particularly speedy SoC it runs on) it’s not
really problematic.
Wow. someone has a real use for it!
I just am too old to become proficient in another language, I am still battling JavaScript
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
GNU has published the definitive version of Hello World:
<https://www.gnu.org/software/hello/>
Richard Kettlewell <[email protected]d> a écrit :
Stéphane CARPENTIER <[email protected]> writes:
Like, it's not a typed language but start playing with dictionaries
without declaring your variables first to see how the claims stand
before the facts.
Python is absolutely a typed language. Every object has a type and it
won’t let you use objects in a manner incompatible with their type:
>>> 0+""
Traceback (most recent call last):
File "<python-input-0>", line 1, in <module>
0+""
~^~~
TypeError: unsupported operand type(s) for +: 'int' and 'str'
In this case python is a strongly typed language and in other cases
it's not that important.
With some typed languages, you can't use a variable if you didn't
declare it's type before. With python, you don't need to declare
integers and strings but you need to declare dictionaries. Why? Don't
know.
With other programming languages I saw, either that works or that
doesn't, but I never saw something that stops working when it starts to
be more complex. It looks like it's a python specialty.
Le 16-10-2026, Chris Ahlstrom <[email protected]> a écrit :
Lawrence D’Oliveiro wrote this screed in ALL-CAPS (fixed):
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
Well C++ is a crime against humanity 'operator overloading...' crap
bloated crap.
Operator overloading is a very handy feature. Yes, C++ is overly
complicated. Java left out that feature of C++, and it, too, still
manages to be excessively complicated.
C++ overly complicated? Nah.
It's so awful it made its creator declare: "There are only two kinds of languages: the ones people complain about and the ones nobody uses."
On 17/10/2026 15:02, Richard Kettlewell wrote:
We make extensive use of Python, apart from some performance issues
(exacerbated by the not particularly speedy SoC it runs on) it’s not
really problematic.
Wow. someone has a real use for it!
I just am too old to become proficient in another language, I am still battling JavaScript
The Natural Philosopher <[email protected]d> writes:
On 17/10/2026 15:02, Richard Kettlewell wrote:
We make extensive use of Python, apart from some performance issues
(exacerbated by the not particularly speedy SoC it runs on) it’s not
really problematic.
Wow. someone has a real use for it!
It is very widely used - most (possibly all?) candidates I’ve
interviewed in the last ten years have at least some background with it (albeit quite often relatively thin experience).
Still, typed doesn’t mean complete immiscibility of types (at least in Python), and a good thing too, since lots of operations take operands of different types. For example if d is an int and G is a Python object representing a point on an elliptic curve, I want to be able to write
d*G,
not mess around with function calls to do point multiplication.
OK I never got much into Python Never needed it for the things I do.
Also it seems to be changing all the time?
But I come from a hardware background, in the sixties there were no 'computers'
Not much memory compared to today's ...
I still have a slide rule somewhere... And old calculators..
https://en.wikipedia.org/wiki/Apollo_Guidance_Computer
Not much memory compared to today's ...
I still have a slide rule somewhere...
<https://www.gnu.org/software/hello/>
Holy cow.
On Sat, 17 Oct 2026 17:12:26 +0100, Richard Kettlewell wrote:
Still, typed doesn’t mean complete immiscibility of types (at least in
Python), and a good thing too, since lots of operations take operands of
different types. For example if d is an int and G is a Python object
representing a point on an elliptic curve, I want to be able to write
d*G,
not mess around with function calls to do point multiplication.
I've been trying to get up to speed with type hints. Like 'private'
functions in Python they seem to depend on a gentleman's agreement not to
do bad things. Like TypeScript it helps but still has loopholes.
(I would also shoot a BB straight up to see how close to me I could get
it to land. A couple of times me and a friend had BB duels.)
On Sat, 17 Oct 2026 18:45:57 GMT, Charlie Gibbs wrote:
<https://www.gnu.org/software/hello/>
Holy cow.
On a similar note, the original Unix “true” command was ... an empty shell script. Nevertheless, AT&T tried to claim copyright on it <http://trillian.mit.edu/~jc/humor/ATT_Copyright_true.html>.
Worth comparing the GNU coreutils version of “true”
<http://www.theregister.co.uk/2016/02/10/line_break_ep2/>, which is
... 80 lines long. The same code is also used to generate the “false” command. The full source is posted in that article, to make fun of
supposed “software bloat” or something. The writer completely fails to realize a point made in the previous link: that the GNU version uses
less memory and runs faster.
On 2026-10-17, Lawrence D’Oliveiro <[email protected]d> wrote:
On Sat, 17 Oct 2026 18:45:57 GMT, Charlie Gibbs wrote:
<https://www.gnu.org/software/hello/>
Holy cow.
On a similar note, the original Unix “true” command was ... an empty
shell script. Nevertheless, AT&T tried to claim copyright on it
<http://trillian.mit.edu/~jc/humor/ATT_Copyright_true.html>.
Worth comparing the GNU coreutils version of “true” >><http://www.theregister.co.uk/2016/02/10/line_break_ep2/>, which is
... 80 lines long. The same code is also used to generate the “false”
command. The full source is posted in that article, to make fun of
supposed “software bloat” or something. The writer completely fails to >> realize a point made in the previous link: that the GNU version uses
less memory and runs faster.
https://en.wikipedia.ort/wiki/IEFBR14
It describes a do-nothing program that needed maintenance anyway.
Lawrence =?iso-8859-13?q?D=FFOliveiro?= <[email protected]d>wrote:
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
Operator overloading is a very handy feature. Yes, C++ is overly
complicated. Java left out that feature of C++, and it, too, still
manages to be excessively complicated.
Python, on the other hand, includes operator overloading, yet is still
only a small fraction of the complexity of either Java or C++. Such a
feature is useful, for example, in letting you express operations on
mathematical objects (like vectors and matrices) using something close
to the original mathematical notation.
OK
I never got much into Python
Never needed it for the things I do.
Also it seems to be changing all the time?
Most code I write is in C or in asm for Microchips.
asm is extremely powerful
understand the hardware!!!
https://panteltje.nl/panteltje/newsflex/download.html https://panteltje.nl/panteltje/quadcopter/index.html https://panteltje.nl/panteltje/pic/index.html
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
But I come from a hardware background, in the sixties there were no 'computers'
yet they did a manned Moon landing and return.
Now they have nothing to show but driving around the block (earth orbit) with all that computing power.
Leroy H <[email protected]>wrote:
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
Most code I write is in C or in asm for Microchips.
asm is extremely powerful
understand the hardware!!!
It is good to hear someone actually standing up for
C and assembly.
These Linux groups seem to be packed only with Python/C++
jockeys. The poor dupes don't know what they're missing.
Assembly is the most fun of any programming language.
There is none of the uncertainty that slows one down
when using a higher level language. The instructions
just flow effortlessly.
However, I do still have trouble with the shuffle operations
on the SIMD registers. I need to always check the manuals
to see what goes where.
Intel and GCC provide support for so-called "intrinsics" that
allow assembly code to be directly incorporated into C code,
but these, for me, are tedious and not much fun to use.
You are also correct in stating that a thorough knowledge
of hardware is a requisite for assembly programming. After
all, one is dealing directly with digital registers with no
cozy compiler or interpreter in between.
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
GNU has published the definitive version of Hello World:
<https://www.gnu.org/software/hello/>
John Bokma <[email protected]>wrote:
On 17/10/2026 19:26, Leroy H wrote:
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
Most code I write is in C or in asm for Microchips.
asm is extremely powerful
understand the hardware!!!
It is good to hear someone actually standing up for
C and assembly.
These Linux groups seem to be packed only with Python/C++
jockeys. The poor dupes don't know what they're missing.
I have programmed several processor families in the past. I am sure I am
not alone. I have done some work in C, including a graph-to-GIF
converter :-).
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
But it often supports Unicode ;-)
Charlie Gibbs <[email protected]d>wrote:
On 2026-10-17, Leroy H <[email protected]> wrote:
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
GNU has published the definitive version of Hello World:
<https://www.gnu.org/software/hello/>
Holy cow. I guess it's time to resurrect this old chestnut:
Mon Nov 12 20:03:48 1990
Message : #1011969 From: cs161fhn%[email protected]
Group : NET.Funny
Length : 198 words
Subject : Evolution of a programmer
Msg-ID: <[email protected]>
Posted: 13 Nov 90 00:30:05 GMT
Person: Dennis Lou
The Evolution of a programmer.
(Okay, it's late and I'm not thinking straight, but at least it's >original...)
High school/Jr. High
--------------------
10 PRINT "HELLO WORLD"
20 END
First year in college
---------------------
program Hello(input, output);
begin
writeln ('Hello world');
end
Senior year in college
----------------------
(defun hello
(cons 'HELLO (list 'WORLD))))
New professional
----------------
#include <stdio.h>
main (argc,argv)
int argc;
char **argv; {
printf ("Hello World!\n");
rbowman <[email protected]>wrote:
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
OK I never got much into Python Never needed it for the things I do.
Also it seems to be changing all the time?
Except for the transition from 2 to 3 Python has been quite good at >maintaining backward compatibility. It changes as PEPs are implemented but >seldom do the changes break things.
But I come from a hardware background, in the sixties there were no
'computers'
I was under the impression the System 360/30 that ate my FORTRAN IV cards >was a computer of sorts.
Richard Kettlewell wrote:
Still, typed doesn’t mean complete immiscibility of types (at least
in Python), and a good thing too, since lots of operations take
operands of different types. For example if d is an int and G is a
Python object representing a point on an elliptic curve, I want to be
able to write d*G, not mess around with function calls to do point
multiplication.
I've been trying to get up to speed with type hints. Like 'private'
functions in Python they seem to depend on a gentleman's agreement not
to do bad things. Like TypeScript it helps but still has loopholes.
rbowman <[email protected]>wrote:
On Sat, 17 Oct 2026 12:17:29 GMT, Jan Panteltje wrote:
Not much memory compared to today's ...
I still have a slide rule somewhere... And old calculators..
https://en.wikipedia.org/wiki/IBM_System/360_Model_30
I believe RPI's had 32k of core. For FFTs etc you wrote partial products
to tape, rewind, and take another pass. Rinse and repeat. rewind() is
still in the C standard library but it was very literal back then.
On Sat, 17 Oct 2026 12:17:29 GMT, Jan Panteltje wrote:
https://en.wikipedia.org/wiki/Apollo_Guidance_Computer
Not much memory compared to today's ...
Well, they weren’t running a complex GUI ;). You entered “Noun” and “Verb” numbers to select which program to run. E.g. the program for
the lunar landing was “P64”. During the Apollo 11 landing, it crashed, twice.
I still have a slide rule somewhere...
I remember reading that the world’s largest slide-rule maker at the
time, Keufel & Esser, went bust in 1971. Electronic digital
calculators had arrived ... not long after the last Moon landing.
c186282 <[email protected]>wrote:
On 10/17/26 03:20, Jan Panteltje wrote:
Lawrence =?iso-8859-13?q?D=FFOliveiro?= <[email protected]d>wrote:
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
Operator overloading is a very handy feature. Yes, C++ is overly
complicated. Java left out that feature of C++, and it, too, still
manages to be excessively complicated.
Python, on the other hand, includes operator overloading, yet is still
only a small fraction of the complexity of either Java or C++. Such a
feature is useful, for example, in letting you express operations on
mathematical objects (like vectors and matrices) using something close
to the original mathematical notation.
OK
I never got much into Python
DO check it out. It can now do basically EVERYTHING
and is faster than it used to be. P4 is promised to
be even faster and more efficient without changing
the syntax.
I did not use Python2 because I heard that P3 was coming ...
why learn an obsolete dialect ? Took 'em a little while
to port everything over to the new version, but it got
done and then went over and above.
CAN be 'compiled' in a few ways. A couple options
translate to 'C' and then compile that. IS much
faster, though not quite as fast as writing a
native 'C' pgm. However the EASE of writing and
great string-slice functions (stolen from FORTRAN)
make Python really nice. Oh, 95% of the time you
don't NEED top speed for applications. That's old
thinking - EASE/comprehensibility can be far
more important.
Oh, the DOCUMENTATION/How-To stuff for Python
is everywhere and detailed. A zillion solutions
for YOUR particular problem.
Since forever I wrote in 'C' or Pascal. I even
did stuff in "B". However I rarely use 'C'
anymore. Might still rec writing 'servers' in 'C' ...
esp if high-volume ... but that sentiment is
half 'tradition'.
Anyway, learn before 'AI' tries to steal it
all away. CHAT can find/write low/medium
complexity Python code ... but sometimes
it seems more complex/verbose/obscure than
a human would write it. Mostly it just
steals human-writ examples it finds.
DID ask CHAT to write me a basic b-tree
database pgm in FORTRAN the other day.
Kinda/sorta did so ... but didn't include
any load/save code (though DID warn about
that prob). Next - same thing, but in
Modula2 - we'll see what its limits are :-)
Pref M3 ... but almost no Linux/anything
compilers for that - esp that actually
install correctly. Some ancient DOS stuff.
Do have a paper manual for "Turbo Modula-2"
but it's all 8/16 bit and won't run on a lot
of chips these days. Would have to use
DOSBox or something - inefficient.
Never needed it for the things I do.
Also it seems to be changing all the time?
Ummm ... not really. The only "big" - ok
not SO big - change was P2 to P3. Most
P2 programs can be tweaked for P3 pretty
quick. Think there's even a converter app.
Now if by "changing" you mean new functions,
yep, people keep writing them for obvious
reasons. The good bit is it means there are
several paths to whatever you're trying to
do - pick the best for your need.
I did a bunch of video security cam apps.
Sometimes "OpenCV2" will do it all, sometimes
"Pillow" lib is just fine. Can even combine
in some cases.
Most code I write is in C or in asm for Microchips.
asm is extremely powerful
Well aware. When CP/M and even DOS were around
had to write ASM bits, or full apps, because
there just wasn't anything better. PICs and
similar microcontrollers - ASM is still the
best way to the finer chip tricks.
understand the hardware!!!
The weaker the hardware the more you have to
understand it. Most modern big chips, so much
overkill that you really don't need to know
every little nuance. Sorry, 1% better performance
isn't worth learning 99% more tiny-tweak shit
that'll probably be obsolete in the very next
point-revision of the CPU.
https://panteltje.nl/panteltje/newsflex/download.html
https://panteltje.nl/panteltje/quadcopter/index.html
https://panteltje.nl/panteltje/pic/index.html
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
But I come from a hardware background, in the sixties there were no 'computers'
yet they did a manned Moon landing and return.
Some of that shit was just bizarre !
Loved hand-knitted "rope memory", what
the LEM computer used :-)
Now they have nothing to show but driving around the block (earth orbit) with
all that computing power.
All the computing power in the Apollo ... yer
base iWatch has a good 1000x that, maybe 10000x.
BUT, they had to make do with what they had ...
and most of that was govt-specced in '64. At
least Apollo wasn't all vacuum tubes :-)
Now space probes - I really admire the people
who design the hardware/software. Insane levels
of redundancy/work-arounds, almost a black art.
It's why the Voyagers are still working out
beyond the edge of our solar system all this
time later. Kinda hoping aliens will find one
and upgrade/repair :-)
On Sat, 17 Oct 2026 18:45:57 GMT, Charlie Gibbs wrote:
<https://www.gnu.org/software/hello/>
Holy cow.
On a similar note, the original Unix “true” command was ... an empty shell script. Nevertheless, AT&T tried to claim copyright on it <http://trillian.mit.edu/~jc/humor/ATT_Copyright_true.html>.
Worth comparing the GNU coreutils version of “true” <http://www.theregister.co.uk/2016/02/10/line_break_ep2/>, which is
... 80 lines long. The same code is also used to generate the “false” command. The full source is posted in that article, to make fun of
supposed “software bloat” or something. The writer completely fails to realize a point made in the previous link: that the GNU version uses
less memory and runs faster.
c186282 <[email protected]>wrote:
On 10/17/26 03:20, Jan Panteltje wrote:
Lawrence =?iso-8859-13?q?D=FFOliveiro?= <[email protected]d>wrote:
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
Operator overloading is a very handy feature. Yes, C++ is overly
complicated. Java left out that feature of C++, and it, too, still
manages to be excessively complicated.
Python, on the other hand, includes operator overloading, yet is still >>>> only a small fraction of the complexity of either Java or C++. Such a
feature is useful, for example, in letting you express operations on
mathematical objects (like vectors and matrices) using something close >>>> to the original mathematical notation.
OK
I never got much into Python
DO check it out. It can now do basically EVERYTHING
and is faster than it used to be. P4 is promised to
be even faster and more efficient without changing
the syntax.
I did not use Python2 because I heard that P3 was coming ...
why learn an obsolete dialect ? Took 'em a little while
to port everything over to the new version, but it got
done and then went over and above.
CAN be 'compiled' in a few ways. A couple options
translate to 'C' and then compile that. IS much
faster, though not quite as fast as writing a
native 'C' pgm. However the EASE of writing and
great string-slice functions (stolen from FORTRAN)
make Python really nice. Oh, 95% of the time you
don't NEED top speed for applications. That's old
thinking - EASE/comprehensibility can be far
more important.
Oh, the DOCUMENTATION/How-To stuff for Python
is everywhere and detailed. A zillion solutions
for YOUR particular problem.
Since forever I wrote in 'C' or Pascal. I even
did stuff in "B". However I rarely use 'C'
anymore. Might still rec writing 'servers' in 'C' ...
esp if high-volume ... but that sentiment is
half 'tradition'.
Anyway, learn before 'AI' tries to steal it
all away. CHAT can find/write low/medium
complexity Python code ... but sometimes
it seems more complex/verbose/obscure than
a human would write it. Mostly it just
steals human-writ examples it finds.
DID ask CHAT to write me a basic b-tree
database pgm in FORTRAN the other day.
Kinda/sorta did so ... but didn't include
any load/save code (though DID warn about
that prob). Next - same thing, but in
Modula2 - we'll see what its limits are :-)
Pref M3 ... but almost no Linux/anything
compilers for that - esp that actually
install correctly. Some ancient DOS stuff.
Do have a paper manual for "Turbo Modula-2"
but it's all 8/16 bit and won't run on a lot
of chips these days. Would have to use
DOSBox or something - inefficient.
Never needed it for the things I do.
Also it seems to be changing all the time?
Ummm ... not really. The only "big" - ok
not SO big - change was P2 to P3. Most
P2 programs can be tweaked for P3 pretty
quick. Think there's even a converter app.
Now if by "changing" you mean new functions,
yep, people keep writing them for obvious
reasons. The good bit is it means there are
several paths to whatever you're trying to
do - pick the best for your need.
I did a bunch of video security cam apps.
Sometimes "OpenCV2" will do it all, sometimes
"Pillow" lib is just fine. Can even combine
in some cases.
Most code I write is in C or in asm for Microchips.
asm is extremely powerful
Well aware. When CP/M and even DOS were around
had to write ASM bits, or full apps, because
there just wasn't anything better. PICs and
similar microcontrollers - ASM is still the
best way to the finer chip tricks.
understand the hardware!!!
The weaker the hardware the more you have to
understand it. Most modern big chips, so much
overkill that you really don't need to know
every little nuance. Sorry, 1% better performance
isn't worth learning 99% more tiny-tweak shit
that'll probably be obsolete in the very next
point-revision of the CPU.
https://panteltje.nl/panteltje/newsflex/download.html
https://panteltje.nl/panteltje/quadcopter/index.html
https://panteltje.nl/panteltje/pic/index.html
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
But I come from a hardware background, in the sixties there were no 'computers'
yet they did a manned Moon landing and return.
Some of that shit was just bizarre !
Loved hand-knitted "rope memory", what
the LEM computer used :-)
Now they have nothing to show but driving around the block (earth orbit) with
all that computing power.
All the computing power in the Apollo ... yer
base iWatch has a good 1000x that, maybe 10000x.
BUT, they had to make do with what they had ...
and most of that was govt-specced in '64. At
least Apollo wasn't all vacuum tubes :-)
Now space probes - I really admire the people
who design the hardware/software. Insane levels
of redundancy/work-arounds, almost a black art.
It's why the Voyagers are still working out
beyond the edge of our solar system all this
time later. Kinda hoping aliens will find one
and upgrade/repair :-)
Yes and the new telescopes
https://www.sciencedaily.com/releases/2026/10/261016070630.htm
NASA’s Roman Space Telescope exceeds expectations by nearly 30%
I think, our current understanding of 'the Universe' and 'big bang' is at some
point completely wrong,
as is our 'understanding' of fusion (why ITER and other experiments like it will never work).
I look at normal radioactivity as you having a lot of springs each compressed and hold together by a little metal clamp.
Every now and then a clamp breaks an the spring jumps, 'radioactivity' 'radiation'.
Put enough springs like that together and if one goes and hits a few other ones
like it then you get a chain reaction, all clamps break, all springs jump.
But 'fusion' they try, is nothing but pushing the springs together without even
using clamps..
Costs all of energy to keep those together and no more energy will ever come out than you put in.
The springs with clamps are just parts of the expanding universe that had not expanded yet.
Dark matter, so much bullshit, so many einStone parrots,
so many mamaticians doing a divide by zero and then use 'renormalisation'
all parrots, too much senseless math, NO understanding.
It is a bit like Ohm's law, but then the radio tube was invented, a current in
a vacuum ! Ohms law broken?
Electrons moving turned out to be what electric current was!
You need a _mechanism_ to understand the universe, else the babble about infinities will go on and no end and the understanding will remain zero.
CERN did recently find signs that space is filled with some fluid...
From:
https://www.sciencedaily.com/releases/2026/10/261004002556.htm
Physicists just saw quarks make waves in the Big Bang’s primordial soup Scientists have watched speeding quarks create ripples in primordial matter, revealing just how liquid-like the early universe really was.
Date:
October 6, 2026
Source:
Massachusetts Institute of Technology
Summary:
Physicists at CERN’s Large Hadron Collider have captured the clearest evidence
yet that quark-gluon plasma, the scorching-hot matter that filled the newborn
universe, behaves like a true liquid.
By tracking individual quarks as they blasted through the plasma, researchers
saw them leave behind ripples, splashes, and swirling wakes much like a duck
moving through water.
The finding suggests this primordial “soup” is so dense that it can slow
speeding quarks and respond collectively as a fluid.
On 17/10/2026 18:46, Stéphane CARPENTIER wrote:
[..]
With some typed languages, you can't use a variable if you didn't
declare it's type before. With python, you don't need to declare
integers and strings but you need to declare dictionaries. Why? Don't
know.
I guess you confuse assigning a type with creating an object of a
specific type.
With python, that looks easy and when I fix something, it breaks
something else for reasons I don't understand.
I recommend to install Copilot CLI on your system and ask it to explain
why your script breaks. It's quite good at explaining is my experience.
Stéphane CARPENTIER <[email protected]> writes:
Richard Kettlewell <[email protected]d> a écrit :
Stéphane CARPENTIER <[email protected]> writes:
Like, it's not a typed language but start playing with dictionaries
without declaring your variables first to see how the claims stand
before the facts.
Python is absolutely a typed language. Every object has a type and it
won’t let you use objects in a manner incompatible with their type:
>>> 0+""
Traceback (most recent call last):
File "<python-input-0>", line 1, in <module>
0+""
~^~~
TypeError: unsupported operand type(s) for +: 'int' and 'str'
In this case python is a strongly typed language and in other cases
it's not that important.
With some typed languages, you can't use a variable if you didn't
declare it's type before. With python, you don't need to declare
integers and strings but you need to declare dictionaries. Why? Don't
know.
No special declaration is needed for dicts, you can just write out the
dict you want.
>>> d={'a': 1, 'b': 2}
>>> type(d)
<class 'dict'>
With other programming languages I saw, either that works or that
doesn't, but I never saw something that stops working when it starts to
be more complex. It looks like it's a python specialty.
No, I think the issue here is pretty obviously the attempt to use a tool without learning how it works.
rbowman <[email protected]>wrote:
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
OK I never got much into Python Never needed it for the things I do.
Also it seems to be changing all the time?
Except for the transition from 2 to 3 Python has been quite good at
maintaining backward compatibility. It changes as PEPs are implemented but >> seldom do the changes break things.
But I come from a hardware background, in the sixties there were no
'computers'
I was under the impression the System 360/30 that ate my FORTRAN IV cards
was a computer of sorts.
Yea I dunno, my first computer was a Sinclair ZX80
At work we had a PDP10?
Worked a while at a large accelerator back then, made me buy a book on Unix. A bit later I had my own TV repair shop in Amsterdam
and that is when I bought that Z80.
Added some printer driver in Z80 asm to print receipts and stuff on a thermal printer.
But then ..expanded that ZX80 with more memory, several I/O cards
even added an IBM PC keyboard and ran the C80 C compiler on it
I wanted to run software from the CPM user group but no CPM for that ZX80.
So wrote my own CP/M clone and it worked:
https://panteltje.nl/panteltje/z80/index.html
Here what I added to that Sinclair Z80:
https://panteltje.nl/panteltje/z80/system14/diagrams/index.html
By that time I was working in a company designing a.o. PCI cards for in IBM PCs
My computer used a ramdisk and was faster than the IBM PC,
no harddisc seek times.
Leroy H wrote this screed in ALL-CAPS (fixed):
It is amazing how knowledge of basic CS is absent in most
programmers, both professional and amateur alike. While
"vibe coding" is all the rage, a silent disaster is also
creeping.
What GNU/Linux requires urgently is a Decimal Floating
Point library that implements IEEE754-2008, but mention this
to most programmers and one will receive only blank stares
and gaping mouths.
The only GNU/Linux project for IEEE754-2008 is closing
its doors:
<https://github.com/libdfp/libdfp/pull/238>
For C++, Boost.Decimal.
The Free42 calculator provides either binary or a decimal
implementation.
By the way, are you AI?
Examples:
Explain as a professional Python programmer why 1+True yields 2 and is
valid
Thought for 3s
In Python, bool is a subclass of int . The two Boolean values behave numerically as False == 0 and True == 1 , so adding True to 1 is
equivalent to 1 + 1 :
1 + True # 2
type(1 + True) # int
It’s valid because Python’s numeric operations accept Boolean values as integers. In application code, though, it’s usually clearer to convert
or count explicitly if you intend to use a Boolean as a number.
On 2026-10-17, The Natural Philosopher <[email protected]d> wrote:
On 17/10/2026 15:02, Richard Kettlewell wrote:
We make extensive use of Python, apart from some performance issues
(exacerbated by the not particularly speedy SoC it runs on) it’s not
really problematic.
Wow. someone has a real use for it!
I just am too old to become proficient in another language, I am still
battling JavaScript
I still use C. I started puttering with Python once -
maybe I should take another look at it.
I think the issue here is pretty obviously the attempt to use a tool
without learning how it works.
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
OK I never got much into Python Never needed it for the things I do.
Also it seems to be changing all the time?
Except for the transition from 2 to 3 Python has been quite good at maintaining backward compatibility. It changes as PEPs are implemented but seldom do the changes break things.
But I come from a hardware background, in the sixties there were no
'computers'
I was under the impression the System 360/30 that ate my FORTRAN IV cards
was a computer of sorts.
Charlie Gibbs <[email protected]d>wrote:
On 2026-10-17, Leroy H <[email protected]> wrote:
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
GNU has published the definitive version of Hello World:
<https://www.gnu.org/software/hello/>
Holy cow. I guess it's time to resurrect this old chestnut:
Mon Nov 12 20:03:48 1990
Message : #1011969 From: cs161fhn%[email protected]
Group : NET.Funny
Length : 198 words
Subject : Evolution of a programmer
Msg-ID: <[email protected]>
Posted: 13 Nov 90 00:30:05 GMT
Person: Dennis Lou
The Evolution of a programmer.
(Okay, it's late and I'm not thinking straight, but at least it's
original...)
High school/Jr. High
--------------------
10 PRINT "HELLO WORLD"
20 END
First year in college
---------------------
program Hello(input, output);
begin
writeln ('Hello world');
end
Senior year in college
----------------------
(defun hello
(cons 'HELLO (list 'WORLD))))
New professional
----------------
#include <stdio.h>
main (argc,argv)
int argc;
char **argv; {
printf ("Hello World!\n");
I will go with that one,
but simplify it a bit
file q1.c
#include <stdio.h>
int main()
{
printf("Hello\n");
}
raspberrypi: ~ # gcc -o q1 q1.c
raspberrypi: ~ # l q1
-rwxr-xr-x 1 root root 8068 Oct 18 09:33 q1*
raspberrypi: ~ # ./q1
Hello
So an about 8 kilobytes size executable ....
Better use bash
echo "Hello"
raspberrypi: ~ # echo "Hello"
Hello
Put enough springs like that together and if one goes and hits a few other ones
like it then you get a chain reaction, all clamps break, all springs jump.
But 'fusion' they try, is nothing but pushing the springs together without even
using clamps..
Costs all of energy to keep those together and no more energy will ever come out than you put in.
The Natural Philosopher <[email protected]d>wrote:
On 18/10/2026 09:53, Jan Panteltje wrote:
Put enough springs like that together and if one goes and hits a few other ones
like it then you get a chain reaction, all clamps break, all springs jump. >>
But 'fusion' they try, is nothing but pushing the springs together without even
using clamps..
Costs all of energy to keep those together and no more energy will ever come >> out than you put in.
Not really. Push enough together hard enough and lots of pieces fly out. >Like trying to assemble a motorbike gearbox with C clips
In the UK we call these 'pingfuckits'
c186282 <[email protected]>wrote:
On 10/18/26 04:53, Jan Panteltje wrote:
c186282 <[email protected]>wrote:
On 10/17/26 03:20, Jan Panteltje wrote:
Lawrence =?iso-8859-13?q?D=FFOliveiro?= <[email protected]d>wrote:
On Fri, 16 Oct 2026 17:10:18 GMT, Jan Panteltje wrote:
Well C++ is a crime against humanity
'operator overloading...' crap bloated crap.
Operator overloading is a very handy feature. Yes, C++ is overly
complicated. Java left out that feature of C++, and it, too, still
manages to be excessively complicated.
Python, on the other hand, includes operator overloading, yet is still >>>>> only a small fraction of the complexity of either Java or C++. Such a >>>>> feature is useful, for example, in letting you express operations on >>>>> mathematical objects (like vectors and matrices) using something close >>>>> to the original mathematical notation.
OK
I never got much into Python
DO check it out. It can now do basically EVERYTHING
and is faster than it used to be. P4 is promised to
be even faster and more efficient without changing
the syntax.
I did not use Python2 because I heard that P3 was coming ...
why learn an obsolete dialect ? Took 'em a little while
to port everything over to the new version, but it got
done and then went over and above.
CAN be 'compiled' in a few ways. A couple options
translate to 'C' and then compile that. IS much
faster, though not quite as fast as writing a
native 'C' pgm. However the EASE of writing and
great string-slice functions (stolen from FORTRAN)
make Python really nice. Oh, 95% of the time you
don't NEED top speed for applications. That's old
thinking - EASE/comprehensibility can be far
more important.
Oh, the DOCUMENTATION/How-To stuff for Python
is everywhere and detailed. A zillion solutions
for YOUR particular problem.
Since forever I wrote in 'C' or Pascal. I even
did stuff in "B". However I rarely use 'C'
anymore. Might still rec writing 'servers' in 'C' ...
esp if high-volume ... but that sentiment is
half 'tradition'.
Anyway, learn before 'AI' tries to steal it
all away. CHAT can find/write low/medium
complexity Python code ... but sometimes
it seems more complex/verbose/obscure than
a human would write it. Mostly it just
steals human-writ examples it finds.
DID ask CHAT to write me a basic b-tree
database pgm in FORTRAN the other day.
Kinda/sorta did so ... but didn't include
any load/save code (though DID warn about
that prob). Next - same thing, but in
Modula2 - we'll see what its limits are :-)
Pref M3 ... but almost no Linux/anything
compilers for that - esp that actually
install correctly. Some ancient DOS stuff.
Do have a paper manual for "Turbo Modula-2"
but it's all 8/16 bit and won't run on a lot
of chips these days. Would have to use
DOSBox or something - inefficient.
Never needed it for the things I do.
Also it seems to be changing all the time?
Ummm ... not really. The only "big" - ok
not SO big - change was P2 to P3. Most
P2 programs can be tweaked for P3 pretty
quick. Think there's even a converter app.
Now if by "changing" you mean new functions,
yep, people keep writing them for obvious
reasons. The good bit is it means there are
several paths to whatever you're trying to
do - pick the best for your need.
I did a bunch of video security cam apps.
Sometimes "OpenCV2" will do it all, sometimes
"Pillow" lib is just fine. Can even combine
in some cases.
Most code I write is in C or in asm for Microchips.
asm is extremely powerful
Well aware. When CP/M and even DOS were around
had to write ASM bits, or full apps, because
there just wasn't anything better. PICs and
similar microcontrollers - ASM is still the
best way to the finer chip tricks.
understand the hardware!!!
The weaker the hardware the more you have to
understand it. Most modern big chips, so much
overkill that you really don't need to know
every little nuance. Sorry, 1% better performance
isn't worth learning 99% more tiny-tweak shit
that'll probably be obsolete in the very next
point-revision of the CPU.
https://panteltje.nl/panteltje/newsflex/download.html
https://panteltje.nl/panteltje/quadcopter/index.html
https://panteltje.nl/panteltje/pic/index.html
These days giggle bytes seem to be needed to just say: 'Hello world'.
Too many layers of crap.
But I come from a hardware background, in the sixties there were no 'computers'
yet they did a manned Moon landing and return.
Some of that shit was just bizarre !
Loved hand-knitted "rope memory", what
the LEM computer used :-)
Now they have nothing to show but driving around the block (earth orbit) with
all that computing power.
All the computing power in the Apollo ... yer
base iWatch has a good 1000x that, maybe 10000x.
BUT, they had to make do with what they had ...
and most of that was govt-specced in '64. At
least Apollo wasn't all vacuum tubes :-)
Now space probes - I really admire the people
who design the hardware/software. Insane levels
of redundancy/work-arounds, almost a black art.
It's why the Voyagers are still working out
beyond the edge of our solar system all this
time later. Kinda hoping aliens will find one
and upgrade/repair :-)
Yes and the new telescopes
https://www.sciencedaily.com/releases/2026/10/261016070630.htm
NASA’s Roman Space Telescope exceeds expectations by nearly 30%
I think, our current understanding of 'the Universe' and 'big bang' is at some
point completely wrong,
Likely.
as is our 'understanding' of fusion (why ITER and other experiments like it >> will never work).
Current approaches ... won't work worth damn.
Gotta find some 'quantum tricks'. SOME way
to, even for a sub-nanosecond, to screw with
the physics.
I look at normal radioactivity as you having a lot of springs each compressed
and hold together by a little metal clamp.
Every now and then a clamp breaks an the spring jumps, 'radioactivity'
'radiation'.
Put enough springs like that together and if one goes and hits a few other ones
like it then you get a chain reaction, all clamps break, all springs jump. >>
But 'fusion' they try, is nothing but pushing the springs together without even
using clamps..
Costs all of energy to keep those together and no more energy will ever come >> out than you put in.
The springs with clamps are just parts of the expanding universe that had not
expanded yet.
Dark matter, so much bullshit, so many einStone parrots,
so many mamaticians doing a divide by zero and then use 'renormalisation'
all parrots, too much senseless math, NO understanding.
It is a bit like Ohm's law, but then the radio tube was invented, a current in
a vacuum ! Ohms law broken?
Electrons moving turned out to be what electric current was!
You need a _mechanism_ to understand the universe, else the babble about
infinities will go on and no end and the understanding will remain zero.
CERN did recently find signs that space is filled with some fluid...
From:
https://www.sciencedaily.com/releases/2026/10/261004002556.htm
Physicists just saw quarks make waves in the Big Bang’s primordial soup
Scientists have watched speeding quarks create ripples in primordial matter, >> revealing just how liquid-like the early universe really was.
Date:
October 6, 2026
Source:
Massachusetts Institute of Technology
Summary:
Physicists at CERN’s Large Hadron Collider have captured the clearest evidence
yet that quark-gluon plasma, the scorching-hot matter that filled the newborn
universe, behaves like a true liquid.
By tracking individual quarks as they blasted through the plasma, researchers
saw them leave behind ripples, splashes, and swirling wakes much like a duck
moving through water.
The finding suggests this primordial “soup” is so dense that it can slow
speeding quarks and respond collectively as a fluid.
Still think "dark matter" is an hallucination.
"Invisible Stuff" ??? Luminiferous Aether ???
Something ELSE going on.
c186282 <[email protected]>wrote:
On 10/18/26 04:01, Jan Panteltje wrote:
rbowman <[email protected]>wrote:
On Sat, 17 Oct 2026 07:20:49 GMT, Jan Panteltje wrote:
OK I never got much into Python Never needed it for the things I do.
Also it seems to be changing all the time?
Except for the transition from 2 to 3 Python has been quite good at
maintaining backward compatibility. It changes as PEPs are implemented but >>> seldom do the changes break things.
But I come from a hardware background, in the sixties there were no
'computers'
I was under the impression the System 360/30 that ate my FORTRAN IV cards >>> was a computer of sorts.
Yea I dunno, my first computer was a Sinclair ZX80
HAD one ! May STILL have it - in The Heap somewhere.
Hey, don't cuss the Z-80 ! Damned good chip !
At work we had a PDP10?
Got started, officially, on the PDP-11.
And there was that neat-o newfangled
'C' language .... :-)
Worked a while at a large accelerator back then, made me buy a book on Unix. >> A bit later I had my own TV repair shop in Amsterdam
and that is when I bought that Z80.
Added some printer driver in Z80 asm to print receipts and stuff on a thermal
printer.
Well, the math of particle accelerators is well
beyond my hard-wiring. I'm best as a 'generalist',
bringing a bunch of dis-related threads together,
seeing how disparate 'A' 'B' and 'C' relate. Had
a good 40+ year career with that.
Sometimes knowing a little about a LOT of things
pays off as well as knowing a LOT about ONE thing.
Somebody has to tie it all together.
But then ..expanded that ZX80 with more memory, several I/O cards
even added an IBM PC keyboard and ran the C80 C compiler on it
I wanted to run software from the CPM user group but no CPM for that ZX80. >> So wrote my own CP/M clone and it worked:
https://panteltje.nl/panteltje/z80/index.html
Here what I added to that Sinclair Z80:
https://panteltje.nl/panteltje/z80/system14/diagrams/index.html
By that time I was working in a company designing a.o. PCI cards for in IBM PCs
My computer used a ramdisk and was faster than the IBM PC,
no harddisc seek times.
Remember 'PCI' cards, 'ISA' too, S-100 were
pretty good as well.
Only designed ONE kinda simple PCI card. Sounds
like you went further.
I fell in love with "embedded" - low-powered
chips and minimal hardware. Seemed just 'art'
to do useful things with that. Lots of 'C'
and ASM shit.
Yea I dunno, my first computer was a Sinclair ZX80 At work we had a
PDP10?
Worked a while at a large accelerator back then, made me buy a book on
Unix.
printf ("Hello World!\n");
But 'fusion' they try, is nothing but pushing the springs together
without even using clamps..
Costs all of energy to keep those together and no more energy will
ever come out than you put in.
rbowman <[email protected]>wrote:
On Sun, 18 Oct 2026 08:01:05 GMT, Jan Panteltje wrote:
Yea I dunno, my first computer was a Sinclair ZX80 At work we had a
PDP10?
Worked a while at a large accelerator back then, made me buy a book on
Unix.
I bought the ZX81 kit to see what you got for $80 US, but that was about
15 years after my first FORTRAN IV program. I wasn't interested in >programming until the microcontrollers appeared so the early part of my >career was hardware control circuits. Logic is logic.
Charlie Gibbs wrote:
printf ("Hello World!\n");
puts("Hello World!\n");
or even a bit more nitpicky:
fputs("Hello World!\n", stdout);
Lawrence =?iso-8859-13?q?D=FFOliveiro?= <[email protected]d>wrote:
On Sun, 18 Oct 2026 08:53:53 GMT, Jan Panteltje wrote:
But 'fusion' they try, is nothing but pushing the springs together
without even using clamps..
Costs all of energy to keep those together and no more energy will
ever come out than you put in.
What keeps the stars burning?
Yep. You can do fixed-point/scaled-integer math even in languages with-IMHO one of the biggest programming mistakes is using floating point
when you can do it faster, more accurately, and more easily with
integers.
Some languages offer direct support for “fixed-point” formats,
basically integers with a fixed scale factor, so you can do
calculations that keep a fractional part.
From what I gather, they tend to be difficult to use properly. The
problem seems to be keeping calculations within the permissible
dynamic range, to avoid overflow and underflow.
This is why floating-point tends to be far more popular.
I do remember Fortran, wrote some lines of that code,
but am C or asm 'oriented'.
On Mon, 19 Oct 2026 08:01:14 GMT, Jan Panteltje wrote:
I do remember Fortran, wrote some lines of that code,
but am C or asm 'oriented'.
I've seldom used Fortran although familiarity with it is handy when converting it to C. After trying to use f2c I decided doing it manually
was necessary if I wanted to be able to read the resulting C.
https://en.wikipedia.org/wiki/BDS_C
That was my first exposure to C.
| Sysop: | DaiTengu |
|---|---|
| Location: | Appleton, WI |
| Users: | 1,142 |
| Nodes: | 10 (0 / 10) |
| Uptime: | 94:08:20 |
| Calls: | 14,704 |
| Calls today: | 2 |
| Files: | 186,497 |
| D/L today: |
2,664 files (888M bytes) |
| Messages: | 2,551,683 |
