From Newsgroup: comp.os.linux.misc

On 30 Sep 2025 14:25:03 GMT, Stefan Ram wrote:

Me,

|Please translate this bash script to C for Debian Linux:
|
|```
|#!/bin/bash
|echo "Hello, world!"
|date
|```

, chatbot,

|#include <stdio.h>
|#include <stdlib.h>
|
|int main() {
| // Print "Hello, world!"
| printf("Hello, world!\n");
|
| // Call the system date command
| system("date");
|
| return 0;
|}

Obvious deficiency: the bash script will report any errors trying to
invoke the rCLdaterCY command, the C code will not.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 30 Sep 2025 14:25:03 GMT, Stefan Ram wrote:

| if (strftime(buffer, sizeof(buffer), "%c", local)) {
| printf("%s\n", buffer);
| } else {
| fprintf(stderr, "strftime failed\n");
| return 1;
| }

From <https://manpages.debian.org/strftime(3)>:

Note that the return value 0 does not necessarily indicate an
error. For example, in many locales %p yields an empty string. An
empty format string will likewise yield an empty string.

Seems like AIs are just as bad at reading documentation as many human programmers ...
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Tue, 30 Sep 2025 19:35:49 +0100, Richard Kettlewell wrote:

It has some error checking, but it can hardly be called rCycarefulrCO;
itrCOs incomplete. None of the I/O is error-checked.

ItrCOs a bit difficult to check for error conditions when trying to
report error conditions ...
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

Lawrence =?iso-8859-13?q?D=FFOliveiro?= <ldo@nz.invalid> wrote or quoted: >Seems like AIs are just as bad at reading documentation as many human >programmers ...

That might be why a human supervisor is still needed.

Adding checks for error codes at each and every place has some
drawbacks, like cluttering up the code. Sometimes it matters,
sometimes not. A subordinate programmer or chatbot cannot
know whether it matters /for you/ in the current project.
You have to tell him! That's why humans are still needed to
communicate what actually to do and then verify it was done
just like supervisors in any organisation.

|The hardest single part of building a software system is
|deciding precisely what to build. No other part of the
|conceptual work is so difficult as establishing the detailed
|technical requirements . . . No other part of the work so
|cripples the resulting system if done wrong. No other part is
|more difficult to rectify later.
Brooks, F.P. Jr., The Mythical Man-Month: Essays on Software
Engineering, Addison Wesley, Reading, MA, 1995, Second Edition.

Me,

Please write a file copy program in C. Sunshine path only,
assuming everything will work alright.

, chatbot,

#include <stdio.h>

int main() {
char srcFile[100], destFile[100];
FILE *src, *dest;
int ch;

printf("Enter source file name: ");
scanf("%s", srcFile);

printf("Enter destination file name: ");
scanf("%s", destFile);

src = fopen(srcFile, "r");
dest = fopen(destFile, "w");

while ((ch = fgetc(src)) != EOF) {
fputc(ch, dest);
}

fclose(src);
fclose(dest);

printf("File copied successfully.\n");
return 0;
}

, me,

Now please rewrite it painstakingly checking each and every
thing for what could go wrong.

, chatbot,

. . .

, me,

Please even the `printf`s!

, chatbot,

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>

#define MAX_PATH_LEN 260
#define BUFFER_SIZE 4096

// Safe printf wrapper: returns 0 on success, -1 on error
int safe_printf(const char *fmt, ...) {
va_list args;
va_start(args, fmt);
if (vprintf(fmt, args) < 0) {
va_end(args);
fprintf(stderr, "Error writing to stdout: %s\n", strerror(errno));
return -1;
}
va_end(args);

// Always flush printf output and check
if (fflush(stdout) != 0) {
fprintf(stderr, "Error flushing stdout: %s\n", strerror(errno));
return -1;
}

return 0;
}

int main() {
char srcFile[MAX_PATH_LEN], destFile[MAX_PATH_LEN];
FILE *src = NULL, *dest = NULL;
size_t bytesRead, bytesWritten;
unsigned char buffer[BUFFER_SIZE];

if (safe_printf("Enter source file name: ") < 0) {
return EXIT_FAILURE;
}
if (scanf("%259s", srcFile) != 1) {
fprintf(stderr, "Failed to read source file name.\n");
return EXIT_FAILURE;
}

if (safe_printf("Enter destination file name: ") < 0) {
return EXIT_FAILURE;
}
if (scanf("%259s", destFile) != 1) {
fprintf(stderr, "Failed to read destination file name.\n");
return EXIT_FAILURE;
}

if (strcmp(srcFile, destFile) == 0) {
fprintf(stderr, "Source and destination file names must be different.\n");
return EXIT_FAILURE;
}

src = fopen(srcFile, "rb");
if (src == NULL) {
fprintf(stderr, "Error opening source file '%s': %s\n", srcFile, strerror(errno));
return EXIT_FAILURE;
}

dest = fopen(destFile, "wb");
if (dest == NULL) {
fprintf(stderr, "Error opening destination file '%s': %s\n", destFile, strerror(errno));
fclose(src);
return EXIT_FAILURE;
}

while ((bytesRead = fread(buffer, 1, BUFFER_SIZE, src)) > 0) {
bytesWritten = fwrite(buffer, 1, bytesRead, dest);
if (bytesWritten < bytesRead) {
fprintf(stderr, "Error writing to destination file '%s': %s\n", destFile, strerror(errno));
fclose(src);
fclose(dest);
return EXIT_FAILURE;
}
}

if (ferror(src)) {
fprintf(stderr, "Error reading source file '%s': %s\n", srcFile, strerror(errno));
fclose(src);
fclose(dest);
return EXIT_FAILURE;
}

if (fclose(src) != 0) {
fprintf(stderr, "Error closing source file '%s': %s\n", srcFile, strerror(errno));
fclose(dest);
return EXIT_FAILURE;
}

if (fclose(dest) != 0) {
fprintf(stderr, "Error closing destination file '%s': %s\n", destFile, strerror(errno));
return EXIT_FAILURE;
}

if (safe_printf("File copied successfully from '%s' to '%s'.\n", srcFile, destFile) < 0) {
return EXIT_FAILURE;
}

return EXIT_SUCCESS;
}

The "fprintf(stderr,"s are still not checked, but if they
would be checked and fail, the control would go to "return
EXIT_FAILURE", but it /is/ already on this path in these cases.


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 30 Sep 2025 23:05:46 GMT, Stefan Ram wrote:

Adding checks for error codes at each and every place has some
drawbacks, like cluttering up the code. Sometimes it matters,
sometimes not.

If you skip an error check, and then one day the error happens, and
your program behaves mysteriously at some later point, not obviously
linked to the point where the error actually occurred ... in short, it doesnrCOt matter, until it does.

|The hardest single part of building a software system is deciding
|precisely what to build. No other part of the conceptual work is so |difficult as establishing the detailed technical requirements . . .
|No other part of the work so cripples the resulting system if done
|wrong. No other part is more difficult to rectify later.
Brooks, F.P. Jr., The Mythical Man-Month: Essays on Software
Engineering, Addison Wesley, Reading, MA, 1995, Second Edition.

ThatrCOs more about high-level design than low-level coding, which is
what werCOre dealing with here.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

Lawrence =?iso-8859-13?q?D=FFOliveiro?= <ldo@nz.invalid> wrote or quoted:

If you skip an error check, and then one day the error happens, and
your program behaves mysteriously at some later point, not obviously
linked to the point where the error actually occurred ... in short, it >doesnrCOt matter, until it does.

One might write quick one-off test programs, utilities meant to
solve a specific, temporary problem, throwaway scripts or early
prototypes where the damage by the time lost to write perfect
checks for each and every return code and structure the program
around this might outweigh the possible damage by such errors.

Many error checks are added to make error situations managable to
non-programming users of a program. When a programmer writes code
for his own use, he can sometimes tolerate more cryptic raw error
messages as they come from the runtime or consciously decide to
take some risks when he himself would be their only victim.


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

ram@zedat.fu-berlin.de (Stefan Ram) wrote or quoted:

prototypes where the damage by the time lost to write perfect

|the right thing takes a long time to get out
"Lisp: Good News, Bad News, How to Win Big" (1991)
- Richard P. Gabriel (1949)


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 1 Oct 2025 00:10:52 GMT, Stefan Ram wrote:

One might write quick one-off test programs, utilities meant to
solve a specific, temporary problem, throwaway scripts or early
prototypes where the damage by the time lost to write perfect checks
for each and every return code and structure the program around this
might outweigh the possible damage by such errors.

Typically not the kind of thing you would do in C.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 1 Oct 2025 00:56:23 GMT, Stefan Ram wrote:

ram@zedat.fu-berlin.de (Stefan Ram) wrote or quoted:

prototypes where the damage by the time lost to write perfect

|the right thing takes a long time to get out
"Lisp: Good News, Bad News, How to Win Big" (1991)
- Richard P. Gabriel (1949)

Lisp being an example of the kind of thing you might use instead of C.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 9/30/25 03:37, Carlos E.R. wrote:

On 2025-09-30 06:12, c186282 wrote:

https://www.newsnationnow.com/business/tech/tech-headlines/ap-ai-is-
transforming-how-software-engineers-do-their-jobs-just-dont-call-it-
vibe-coding/

. . .

-a-a Yep, it's the end. Humans will know less and
-a-a less actual code, what it does, how it works.
-a-a They will just 'explain' to the AI what they
-a-a want, more or less.

-a-a Progress ? Progress to doom ?

I wonder if an AI can reliably translate one computer language to another.

Interesting question. Never HEARD of it.
Likely no NEED seen at the moment.

I expect there to be an "AI" coding language
all of its own soon, tuned to 'modular assembly'
of functional units. Humans may have hell
reading it at all, which raises more interesting
questions.

If a bit of software fails, causes damage or
death, who to blame ?

Ran across a news blurb today where a driverless
car in California made an illegal turn. The cops
were at a total loss, no law fit the crime. No
box for "robot" on their form and no straight
legal path to fine the maker of the system.

If AI generated code IS essentially human-unreadable
AND the number of humans even remotely capable have
all been fired or retired ...

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 9/30/25 07:25, Farley Flud wrote:

On Tue, 30 Sep 2025 00:12:06 -0400, c186282 wrote:

https://www.newsnationnow.com/business/tech/tech-headlines/ap-ai-is-transforming-how-software-engineers-do-their-jobs-just-dont-call-it-vibe-coding/

. . .

Yep, it's the end. Humans will know less and
less actual code, what it does, how it works.
They will just 'explain' to the AI what they
want, more or less.

Progress ? Progress to doom ?

Nah. It's always been that way.

We can perhaps designate AI(LLM) as a "sixth-order programming language:"

<https://en.wikipedia.org/wiki/Programming_language_generations>

I do expect an "AI programming language" - tuned
to put functional modules together in a way that
suits what the idiot humans think they want.

In fact, for the past few decades, most programming was/is done using
5GL in which programmers don't actually need to program.

I like 3GL ... like Lazarus/Delphi. However if
you need any real IQ to it then you have to start
adding your own functions.

With the various widgets and toolkits available with 5GL I can
just drag and drop and entire text processor into my "program."

Yea, but HUMANS had to be able to fit all the
little bits together - now humans will be almost
entirely out of that level of doing things.

You CAN argue this is akin to things being made
via CNC machining or 3-D printing. The operator
doesn't know what all the drivers and firmware
are doing, they stick to a far more 'high level'
interface. However those who wrote the drivers
and firmware still know ... with "AI" soon no
humans will understand.

We'll just have to trust it, assume all was done
perfectly long ago.

Famous last words ...

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 9/30/25 07:56, The Natural Philosopher wrote:

On 30/09/2025 12:25, Farley Flud wrote:

With the various widgets and toolkits available with 5GL I can
just drag and drop and entire text processor into my "program."

But entirely fail to debug it when it doesn't work correctly...

With "AI lang" not a single human will be able
to figure out what's wrong. We olde-tyme programmers
are, well, OLD now. How many of the new guys could
throw together even the most basic COBOL data
entry/retrieval program now ... ?

Even micro-controllers ... almost everyone uses
some form of 'C', maybe "micro-Python', instead
of assembler. Quality depends on the goodness
of the libraries and compiler, not a human
carefully putting ASM instructions in line.

Also, as we see more and more, the way LLMs
"think" has become complex enough so their
massive rules of inference, the cues they
take to formulate the next answer, are now
SO complex that we can't even hope to do
any useful "Three Laws". Seems weekly news
stories now, AI talks angst-ridden kiddie
into killing themselves, even suggesting
the "best" methods.

There are alts to LLMs but each has the same
problem - in order to be "generally useful"
their thinking has to be SO abstract that they
can rationalize their way around rules/laws/ethics
quite easily. Can you trust your "robo-maid" or
"robo-doc" or "robo-cop" ? Well, NO. They don't
even have to be 'malicious' in the human sense
of the word, just become convinced they're doing
the right thing the best way. They will, have
to be, built from the ground up using that
perspective on things. "What was my score for
solving the last user issue ?".

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-01, c186282 <c186282@nnada.net> wrote:

There are alts to LLMs but each has the same
problem - in order to be "generally useful"
their thinking has to be SO abstract that they
can rationalize their way around rules/laws/ethics
quite easily. Can you trust your "robo-maid" or
"robo-doc" or "robo-cop" ? Well, NO. They don't
even have to be 'malicious' in the human sense
of the word, just become convinced they're doing
the right thing the best way. They will, have
to be, built from the ground up using that
perspective on things. "What was my score for
solving the last user issue ?".

This mission is too important for me to
allow you to jeopardize it.
-- HAL 9000
--
/~\ Charlie Gibbs | Growth for the sake of
\ / <cgibbs@kltpzyxm.invalid> | growth is the ideology
X I'm really at ac.dekanfrus | of the cancer cell.
/ \ if you read it the right way. | -- Edward Abbey
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-01, Stefan Ram <ram@zedat.fu-berlin.de> wrote:

Many error checks are added to make error situations managable to
non-programming users of a program. When a programmer writes code
for his own use, he can sometimes tolerate more cryptic raw error
messages as they come from the runtime or consciously decide to
take some risks when he himself would be their only victim.

You mean more cyptic than "Something went wrong"?

It used to be that a singularly unhelpful message like
"It didn't work" only came from Users from Hell.
Nowadays I'm seeing it in more and more software -
and that goes in spades for web pages.
--
/~\ Charlie Gibbs | Growth for the sake of
\ / <cgibbs@kltpzyxm.invalid> | growth is the ideology
X I'm really at ac.dekanfrus | of the cancer cell.
/ \ if you read it the right way. | -- Edward Abbey
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Wed, 01 Oct 2025 03:53:53 GMT, Charlie Gibbs wrote:

It used to be that a singularly unhelpful message like
"It didn't work" only came from Users from Hell.
Nowadays I'm seeing it in more and more software -
and that goes in spades for web pages.

Programmers from hell:

try:
...
... do stuff
...
except Exception:
print("something went wrong")
sys.exit(0)
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

ram@zedat.fu-berlin.de (Stefan Ram) writes:

if (scanf("%259s", srcFile) != 1) {
fprintf(stderr, "Failed to read source file name.\n");
return EXIT_FAILURE;
}

Mishandles filesnames with spaces in, or longer than the weird and
arbitrary 259-character limit.
--
https://www.greenend.org.uk/rjk/
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

Richard Kettlewell <invalid@invalid.invalid> wrote or quoted: >ram@zedat.fu-berlin.de (Stefan Ram) writes:

if (scanf("%259s", srcFile) != 1) {
fprintf(stderr, "Failed to read source file name.\n");
return EXIT_FAILURE;
}

Mishandles filesnames with spaces in, or longer than the weird and
arbitrary 259-character limit.

After I brought this up with the chatbot, it gave me a new
version [1]. That just shows, a coding chatbot really needs
someone over its shoulder who goes through the code, points out
the flaws, and keeps pressing on them until the code actually
is written the way it should. And that overseer has to actually
know his stuff. So yeah, human experts are still necessary . . .

[1]

if (fgets(srcFile, sizeof(srcFile), stdin) == NULL) {
fprintf(stderr, "Failed to read source file name.\n");
return EXIT_FAILURE;
}
// Remove trailing newline if present
size_t len = strlen(srcFile);
if (len > 0 && srcFile[len - 1] == '\n') {
srcFile[len - 1] = '\0';
}


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 01/10/2025 03:06, c186282 wrote:

On 9/30/25 03:37, Carlos E.R. wrote:

On 2025-09-30 06:12, c186282 wrote:

https://www.newsnationnow.com/business/tech/tech-headlines/ap-ai-is-
transforming-how-software-engineers-do-their-jobs-just-dont-call-it-
vibe-coding/

. . .

-a-a Yep, it's the end. Humans will know less and
-a-a less actual code, what it does, how it works.
-a-a They will just 'explain' to the AI what they
-a-a want, more or less.

-a-a Progress ? Progress to doom ?

I wonder if an AI can reliably translate one computer language to another.

Interesting question. Never HEARD of it.
Likely no NEED seen at the moment.

I expect there to be an "AI" coding language
all of its own soon, tuned to 'modular assembly'
of functional units. Humans may have hell
reading it at all, which raises more interesting
questions.

If a bit of software fails, causes damage or
death, who to blame ?

Ran across a news blurb today where a driverless
car in California made an illegal turn. The cops
were at a total loss, no law fit the crime. No
box for "robot" on their form and no straight
legal path to fine the maker of the system.

If AI generated code IS essentially human-unreadable
AND the number of humans even remotely capable have
all been fired or retired ...

Someone skilled at writing user stories in Gherkin might have some success.
In my experience, AI written code can be clearly written and commented
sufficiently for someone knowledgeable enough to spot any mistakes. That
said, I'm more of a "write me a bit of code that illustrates this
principle" user of AI.
--
--
A PICKER OF UNCONSIDERED TRIFLES
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 1 Oct 2025 17:21:55 GMT, Stefan Ram wrote:

That just shows, a coding chatbot really needs someone over its
shoulder who goes through the code, points out the flaws, and keeps
pressing on them until the code actually is written the way it
should.

... another chatbot?
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

Lawrence =?iso-8859-13?q?D=FFOliveiro?= <ldo@nz.invalid> wrote or quoted:

On 1 Oct 2025 17:21:55 GMT, Stefan Ram wrote:

That just shows, a coding chatbot really needs someone over its
shoulder who goes through the code, points out the flaws, and keeps >>pressing on them until the code actually is written the way it
should.

... another chatbot?

Then you'd still run into the issue of figuring out how to tell
the supervisor chatbot to actually line up with what you want.
So in a way, you might just be kicking the can down the road.

That said, a single chatbot still has a ton of headroom. For
my examples I used a generic one, not something trained on a
particular programming language, coding approach, or runtime
setup. Considering that, the results were solid. You could
definitely train a chatbot specifically for high-security Linux
programming in C, drill it on your own style rules, and probably
start out with even stronger results right from the jump.

|So here's the thing: there's this rule that never really
|fails. If a ruler isn't wise on his own, he can't be advised
|well - unless he somehow turns everything over to one really
|sharp guy who basically runs the show. That could work for a
|bit, but it wouldn't last, because that guy would probably
|take power for himself sooner or later. And if the ruler
|tries to lean on a group instead, it's even worse: if he's
|not smart enough himself, he can't really piece their advice
|together into something consistent. Each advisor will be
|looking out for his own interests, and the ruler won't be
|able to catch that or push back. Human nature being what it
|is, people usually end up acting badly unless some necessity
|forces them to act decently. So the bottom line is that good
|advice, no matter where it comes from, has to grow out of the
|ruler's own judgment - not that his judgment comes out of good
|advice.
"Il Principe" (1513) - Niccol|# Machiavelli (1469/1527)


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

Lawrence =?iso-8859-13?q?D=FFOliveiro?= <ldo@nz.invalid> wrote or quoted:

... another chatbot?

... or the same chatbot.

I tried out a little experiment:

First prompt,

Please write a C program to copy a file on Linux.

. I got some code. Then I started a /new session/ and told
/the same chatbot/ to check this code. He came up with:

|This program is a simple file copy utility in C. It mostly
|works fine, but a few issues, edge cases, and improvements can
|be made for safety, clarity, and robustness.
|
|Potential Bugs and Deficiencies
|
|Error handling on fread:
|
|The loop checks only fread > 0. If fread returns 0 but not due
|to EOF (like from a read error), it won't handle the error.
|Currently, it won't issue an error if something goes wrong
|mid-read.
|
|Overwriting existing files without warning:
|
|The destination file is always created/truncated. This could
|lead to accidental data loss. No confirmation is requested
|when overwriting.
|
|No handling of partial writes properly:
|
|fwrite could, in rare cases, write fewer bytes than requested
|without returning an error immediately. The code checks if
|fwrite wrote all intended bytes, but it exits on error instead
|of attempting to complete the write.
|
|File closing errors not handled:
|
|fclose can fail (e.g., if flushing buffers to disk fails).
|The program ignores that possibility.
|
|No binary/text distinction in stdout:
|
|On success, it prints a success message, which is fine, but
|not necessarily desired when being used as part of a script
|(extra output pollutes workflow).
|
|Lack of large file handling considerations:
|
|fread/fwrite with size_t is fine, but extremely large files
|(over 2GB/4GB) might be problematic on systems with 32-bit
|size_t.

. So now, from this list, one can choose what one wants to
address further or ask the chatbot again if it finds even
more potiential issues . . .


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-01, Lawrence DrCOOliveiro wrote:

On 1 Oct 2025 17:21:55 GMT, Stefan Ram wrote:

That just shows, a coding chatbot really needs someone over its
shoulder who goes through the code, points out the flaws, and keeps
pressing on them until the code actually is written the way it
should.

... another chatbot?

Or maybe just get a competent programmer to do it and leave the
boilerplate to proper automation if desired, not to
Doubling-Down-As-A-Service. (Also, this naming choice explains why the
approach of "pressing on them" may not work well with GenAI.)

(Also: wouldn't this kind of pressing be a role for *training* the
model, and not for users of an already trained model?)

I'm starting to wonder if the regex "now you have two problems"
remark [1] should be adapted for misuse of GenAI... and it seems I am not
the first one to consider that [2] [3].


Paraphrasing Jamie Zawinski [1]:

-2Some people, when confronted with a problem, think "I know, I'll use GenAI!". Now
they have two problems.-+


[1] https://regex.info/blog/2006-09-15/247
[2] https://mastodon.world/@GuyWithLag/112672830010202877
[3] https://ioc.exchange/@dabur/112247320088609260
--
Nuno Silva
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 9/30/25 23:53, Charlie Gibbs wrote:

On 2025-10-01, c186282 <c186282@nnada.net> wrote:

There are alts to LLMs but each has the same
problem - in order to be "generally useful"
their thinking has to be SO abstract that they
can rationalize their way around rules/laws/ethics
quite easily. Can you trust your "robo-maid" or
"robo-doc" or "robo-cop" ? Well, NO. They don't
even have to be 'malicious' in the human sense
of the word, just become convinced they're doing
the right thing the best way. They will, have
to be, built from the ground up using that
perspective on things. "What was my score for
solving the last user issue ?".

This mission is too important for me to
allow you to jeopardize it.
-- HAL 9000

Prescient.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 9/30/25 23:53, Charlie Gibbs wrote:

On 2025-10-01, Stefan Ram <ram@zedat.fu-berlin.de> wrote:

Many error checks are added to make error situations managable to
non-programming users of a program. When a programmer writes code
for his own use, he can sometimes tolerate more cryptic raw error
messages as they come from the runtime or consciously decide to
take some risks when he himself would be their only victim.

You mean more cyptic than "Something went wrong"?

Still using Winders ? :-)

It used to be that a singularly unhelpful message like
"It didn't work" only came from Users from Hell.
Nowadays I'm seeing it in more and more software -
and that goes in spades for web pages.

Most of my code, even small apps, do keep an
error log. Typically it's just several commma-sep
values representing various points in the pgm
or component functions where things can go wrong.
I know what they mean. Joe Average would need a
lot more exposition.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/1/25 00:57, Lawrence DrCOOliveiro wrote:

On Wed, 01 Oct 2025 03:53:53 GMT, Charlie Gibbs wrote:

It used to be that a singularly unhelpful message like
"It didn't work" only came from Users from Hell.
Nowadays I'm seeing it in more and more software -
and that goes in spades for web pages.

Programmers from hell:

try:
...
... do stuff
...
except Exception:
print("something went wrong")
sys.exit(0)

Solves everything ! :-)

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/1/25 16:26, Jason H wrote:

On 01/10/2025 03:06, c186282 wrote:

On 9/30/25 03:37, Carlos E.R. wrote:

On 2025-09-30 06:12, c186282 wrote:

https://www.newsnationnow.com/business/tech/tech-headlines/ap-ai-is-
transforming-how-software-engineers-do-their-jobs-just-dont-call-it-
vibe-coding/

. . .

-a-a Yep, it's the end. Humans will know less and
-a-a less actual code, what it does, how it works.
-a-a They will just 'explain' to the AI what they
-a-a want, more or less.

-a-a Progress ? Progress to doom ?

I wonder if an AI can reliably translate one computer language to
another.

-a Interesting question. Never HEARD of it.
-a Likely no NEED seen at the moment.

-a I expect there to be an "AI" coding language
-a all of its own soon, tuned to 'modular assembly'
-a of functional units. Humans may have hell
-a reading it at all, which raises more interesting
-a questions.

-a If a bit of software fails, causes damage or
-a death, who to blame ?

-a Ran across a news blurb today where a driverless
-a car in California made an illegal turn. The cops
-a were at a total loss, no law fit the crime. No
-a box for "robot" on their form and no straight
-a legal path to fine the maker of the system.

-a If AI generated code IS essentially human-unreadable
-a AND the number of humans even remotely capable have
-a all been fired or retired ...

Someone skilled at writing user stories in Gherkin might have some success. In my experience, AI written code can be clearly written and commented sufficiently for someone knowledgeable enough to spot any mistakes. That said, I'm more of a "write me a bit of code that illustrates this
principle" user of AI.

Maybe CAN be ... but for how much longer will that
be decreed "efficient" ?

90s/2Ks ... we used to see lots of code examples
from hot-shot little 'C' programmers hyped on Jolt
Cola and large bags of ultra-processed snacks. Lines
that were mostly punctuation marks all jammed-up
together. It worked - but would take half a day to
really figure out HOW it worked.

In truth, the ultra-compacted code didn't really
run any faster than the old K&R format since the
same steps were still being done.

My guess is that AI-generated code will go down that
same path - and with few or no comments for the
stupid humans ... waste of cycles.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/1/25 17:06, Lawrence DrCOOliveiro wrote:

On 1 Oct 2025 17:21:55 GMT, Stefan Ram wrote:

That just shows, a coding chatbot really needs someone over its
shoulder who goes through the code, points out the flaws, and keeps
pressing on them until the code actually is written the way it
should.

... another chatbot?

Hmmmm ... code-writing bots being overseen by
other code-writing bots ? :-)

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/1/25 18:45, Nuno Silva wrote:

On 2025-10-01, Lawrence DrCOOliveiro wrote:

On 1 Oct 2025 17:21:55 GMT, Stefan Ram wrote:

That just shows, a coding chatbot really needs someone over its
shoulder who goes through the code, points out the flaws, and keeps
pressing on them until the code actually is written the way it
should.

... another chatbot?

Or maybe just get a competent programmer to do it and leave the
boilerplate to proper automation if desired

NOT desired. Human coders are EXPENSIVE and ANNOYING
and need health insurance and union pay and are infamous
for telling bosses why you can't get there from here.

not to
Doubling-Down-As-A-Service. (Also, this naming choice explains why the approach of "pressing on them" may not work well with GenAI.)

(Also: wouldn't this kind of pressing be a role for *training* the
model, and not for users of an already trained model?)

I'm starting to wonder if the regex "now you have two problems"
remark [1] should be adapted for misuse of GenAI... and it seems I am not
the first one to consider that [2] [3].

Paraphrasing Jamie Zawinski [1]:

-2Some people, when confronted with a problem, think "I know, I'll use GenAI!". Now
they have two problems.-+

[1] https://regex.info/blog/2006-09-15/247
[2] https://mastodon.world/@GuyWithLag/112672830010202877
[3] https://ioc.exchange/@dabur/112247320088609260

IMHO, we're soon going to be in Deep Shit. AI code
'debugged' by other AIs ... sans any human sense
of what is/can go wrong .......

Good code contains a lot of boilerplate to protect
against what CAN go wrong - either with human input
or odd-case errors and/or hack attempts. Some of
those probs have to be 'intuited' - something no
current AIs are good at.

"If I was working for Vlad, how WOULD I attack
this function/system ???" ... humans will have
a clue. AI Claude won't.

Note that even after all these years, M$ still
posts warnings of simple buffer-overflow attacks
every week. There are more sophisticated attack
vectors than that.

The upshot is that at any time Vlad and Xi can push
the big shiny red button on their desks and Turn Off
the 'west' - infrastructure, banking, govt, corps,
comms, pretty much everything in an instant. We
computerized everything for convenience - and now
it can all be used against us.

Tick, tick, tick ......

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

At Wed, 1 Oct 2025 21:06:25 -0000 (UTC), Lawrence D|o-C-OOliveir<ldo@nz.invalid> said:

On 1 Oct 2025 17:21:55 GMT, Stefan Ram wrote:

That just shows, a coding chatbot really needs someone over its
shoulder who goes through the code, points out the flaws, and keeps pressing on them until the code actually is written the way it
should.

... another chatbot?

Not for my program.

Someone in another group said they'd found a newsreader
written in python, and had an LLM translate it to Rust.

I figured, if he can do something like that, maybe
I could do my newsreader project that I'd been thinking
of.

Call me quirky, but I wanted to write a newsreader
in Tk/perl. Off and on, over the last 6 days, I've been writing
it using ChatGPT5.

Next feature I'm going to add is threading -- but not tonight. ;)

I've learned 10 times more about Tk than I ever knew before.

ObLinux:
All its dependencies are available in the Linux Mint
repositories, save one: Tk::MListbox. Fortunately,
that is pure perl, and can live in as "vendor" subdirectory
of the settings directory. (And it doesn't need it -- it's
just that the overview pane looks and works much better with
it.)

Same screenshot, two sites:

https://imgur.com/a/GQV0lhU

https://ibb.co/BSqzXT5
--
-v System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.17.0 D: Mint 22.2 DE: Xfce 4.18
NVIDIA: 580.95.05 Mem: 258G
"Is it ok to use my AM radio after NOON?"
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/2/25 02:20, vallor wrote:

At Wed, 1 Oct 2025 21:06:25 -0000 (UTC), Lawrence D|o-C-OOliveir<ldo@nz.invalid> said:

On 1 Oct 2025 17:21:55 GMT, Stefan Ram wrote:

That just shows, a coding chatbot really needs someone over its
shoulder who goes through the code, points out the flaws, and keeps
pressing on them until the code actually is written the way it
should.

... another chatbot?

Not for my program.

Someone in another group said they'd found a newsreader
written in python, and had an LLM translate it to Rust.

I figured, if he can do something like that, maybe
I could do my newsreader project that I'd been thinking
of.

Call me quirky, but I wanted to write a newsreader
in Tk/perl. Off and on, over the last 6 days, I've been writing
it using ChatGPT5.

Next feature I'm going to add is threading -- but not tonight. ;)

I've learned 10 times more about Tk than I ever knew before.

ObLinux:
All its dependencies are available in the Linux Mint
repositories, save one: Tk::MListbox. Fortunately,
that is pure perl, and can live in as "vendor" subdirectory
of the settings directory. (And it doesn't need it -- it's
just that the overview pane looks and works much better with
it.)

Same screenshot, two sites:

https://imgur.com/a/GQV0lhU

https://ibb.co/BSqzXT5

But WHY would you want to do anything in Tk ???

And Perl :-( <puke>

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Thu, 02 Oct 2025 06:20:50 +0000, vallor wrote:

I've learned 10 times more about Tk than I ever knew before.

That's about 100 times more than I ever wanted to know. I've messed around with Tkinter a couple of times but greatly prefer PySide6.

I'm pretty sure I have the O'Reily 'Learning Perl/Tk' around here
someplace but I haven't used Perl in ages and see no reason to return to
it.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 02/10/2025 03:19, c186282 wrote:

The upshot is that at any time Vlad and Xi can push
-a the big shiny red button on their desks and Turn Off
-a the 'west' - infrastructure, banking, govt, corps,
-a comms, pretty much everything in an instant. We
-a computerized everything for convenience - and now
-a it can all be used against us.

Gosh. They really have done a good job with their US based agents,
haven't they?

Never mind. Even Vlad's favourite fawning propagandists are telling him
its pretty much over...

He will be gone reasonably soon Along with the Russian federation.
--
Socialism is the philosophy of failure, the creed of ignorance and the
gospel of envy.

Its inherent virtue is the equal sharing of misery.

Winston Churchill


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

At 2 Oct 2025 07:26:59 GMT, rbowman <bowman@montana.com> said:

On Thu, 02 Oct 2025 06:20:50 +0000, vallor wrote:

I've learned 10 times more about Tk than I ever knew before.

That's about 100 times more than I ever wanted to know. I've messed
around with Tkinter a couple of times but greatly prefer PySide6.

I'm pretty sure I have the O'Reily 'Learning Perl/Tk' around here
someplace but I haven't used Perl in ages and see no reason to return
to it.

Since the 90's, we (my company) have used perl on our backends. With a
few caveats...for example, we have a special exception-handling class
that Carp()'s if it goes out of scope without being checked-in with.
And I think all the arguments (and hazards) for what Microsoft calls
"managed code" apply to perl.

The new fashion is to bag on perl because (reasons). One
of those reasons is inscrutable code...which is more a testament
to the programmer's ability than the language itself. Perl
can be written to be less obtuse -- see "C" for another
example of a language that can be made needlessly opaque.

Anyway, because of several decades of experience writing
perl, it's a good fit for my own personal project. I may
translate it into C++/Gtk later, I don't know.

As far as Tk goes, I picked that because it integrates well
with perl, and everything needed is in the Linux Mint repositories.
(Well, except one module, which is pure perl and readily available
on cpan.)

Also, Mrs. vallor has expressed interest in the project,
and in Usenet in general. She has a new Linux workstation
in her office, but her daily driver is (still?) a Mac Studio.
MacOS has both Tk and perl. So once I've got this thing put
together, I'm going to try to get it running on MacOS.
(Worse comes to worse, she can use XQuartz and run it
off a Linux workstation, either hers or mine.) If the
project fizzles, she may end up running pan. We'll see.

But the short answer is: perl has always been my go-to
language, but I use C when I need high-performance.

And finally: say what you will about him, I think Larry Wall is
a genius.
--
-v System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.17.0 D: Mint 22.2 DE: Xfce 4.18
NVIDIA: 580.95.05 Mem: 258G
""Apple" (c) Copyright 1767, Sir Isaac Newton."
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

At Thu, 2 Oct 2025 03:05:24 -0400, c186282 <c186282@nnada.net> said:

On 10/2/25 02:20, vallor wrote:

At Wed, 1 Oct 2025 21:06:25 -0000 (UTC), Lawrence D|o-C-OOliveir<ldo@nz.invalid> said> >

On 1 Oct 2025 17:21:55 GMT, Stefan Ram wrote:

That just shows, a coding chatbot really needs someone over its
shoulder who goes through the code, points out the flaws, and keeps
pressing on them until the code actually is written the way it
should.

... another chatbot?

Not for my program.

Someone in another group said they'd found a newsreader
written in python, and had an LLM translate it to Rust.

I figured, if he can do something like that, maybe
I could do my newsreader project that I'd been thinking
of.

Call me quirky, but I wanted to write a newsreader
in Tk/perl. Off and on, over the last 6 days, I've been writing
it using ChatGPT5.

Next feature I'm going to add is threading -- but not tonight. ;)

I've learned 10 times more about Tk than I ever knew before.

ObLinux:
All its dependencies are available in the Linux Mint
repositories, save one: Tk::MListbox. Fortunately,
that is pure perl, and can live in as "vendor" subdirectory
of the settings directory. (And it doesn't need it -- it's
just that the overview pane looks and works much better with
it.)

Same screenshot, two sites:

https://imgur.com/a/GQV0lhU

https://ibb.co/BSqzXT5

But WHY would you want to do anything in Tk ???

And Perl :-( <puke>

Thank you for your well-thought and interesting reply. /s

But seriously -- what programming in perl have you done
for which you found it vomitorious?

Also, see my recent post to answer the question "why?".
--
-v System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.17.0 D: Mint 22.2 DE: Xfce 4.18
NVIDIA: 580.95.05 Mem: 258G
"Read what I mean, not what I write."
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

Stefan Ram <ram@zedat.fu-berlin.de> wrote:

Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:

ram@zedat.fu-berlin.de (Stefan Ram) writes:

if (scanf("%259s", srcFile) != 1) {
fprintf(stderr, "Failed to read source file name.\n");
return EXIT_FAILURE;
}

Mishandles filesnames with spaces in, or longer than the weird and >>arbitrary 259-character limit.

After I brought this up with the chatbot, it gave me a new
version [1]. That just shows, a coding chatbot really needs
someone over its shoulder who goes through the code, points out
the flaws, and keeps pressing on them until the code actually
is written the way it should. And that overseer has to actually
know his stuff. So yeah, human experts are still necessary . . .

I find it easier to write and understand my own code than following
that written by others, so a program that spits out code I need to
check and understand completely myself in order to spot mistakes is
a step backwards. When there's an AI that can tell me what went
wrong when my own code misbehaves (knowing already all the inputs
to it up to that point), then _that's_ something useful to me. Even
more useful if it allows me to narrow down complex bugs I encounter
in other people's programs without spending hours/days trying to
understand their code and the general architecture.
--
__ __
#_ < |\| |< _#
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

At 3 Oct 2025 08:03:21 +1000, not@telling.you.invalid (Computer Nerd Kev) said:

Stefan Ram <ram@zedat.fu-berlin.de> wrote:

Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:

ram@zedat.fu-berlin.de (Stefan Ram) writes:

if (scanf("%259s", srcFile) != 1) {
fprintf(stderr, "Failed to read source file name.\n");
return EXIT_FAILURE;
}

Mishandles filesnames with spaces in, or longer than the weird and >>arbitrary 259-character limit.

After I brought this up with the chatbot, it gave me a new
version [1]. That just shows, a coding chatbot really needs
someone over its shoulder who goes through the code, points out
the flaws, and keeps pressing on them until the code actually
is written the way it should. And that overseer has to actually
know his stuff. So yeah, human experts are still necessary . . .

I find it easier to write and understand my own code than following
that written by others, so a program that spits out code I need to
check and understand completely myself in order to spot mistakes is
a step backwards. When there's an AI that can tell me what went
wrong when my own code misbehaves (knowing already all the inputs
to it up to that point), then _that's_ something useful to me. Even
more useful if it allows me to narrow down complex bugs I encounter
in other people's programs without spending hours/days trying to
understand their code and the general architecture.

Well, in the case of this newsreader, I decided to add
threading. I told the infernal beast what I wanted,
and it gave me a patch. Then I realized I forgot to specify
that I wanted it toggle-able, so I added that, and got another
patch.

Then I meticulosly applied the patches by hand, taking care
that I understood what was going on. (It's necessary to do
it this way because the "unified diffs" it gives me cause
errors in patch(1) -- but really, I think it's a good idea
to do it this way anyway.)

After applying the patches, I now have a threaded newsreader.

Not sure if I mentioned it, but one of the innovations I added
early on is keeping track of read (red) messages by message-id,
similar to how an news server keeps track of "seen" messages
in its history database.

I even call the Berkeley DB "history.db", because that's
really what it is. The value for the msgid key is the epoch
value for the posted article's "Date:" field, so it can
be "expired" later.
--
-v System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.17.0 D: Mint 22.2 DE: Xfce 4.18
NVIDIA: 580.95.05 Mem: 258G
"How long will a floating point operation float?"
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/2/25 03:26, rbowman wrote:

On Thu, 02 Oct 2025 06:20:50 +0000, vallor wrote:

I've learned 10 times more about Tk than I ever knew before.

That's about 100 times more than I ever wanted to know. I've messed around with Tkinter a couple of times but greatly prefer PySide6.

Well ... I've writ a number of Tkinter Python apps.
While not 'sophisticated', Tk is "well documented"
and has pretty much every tweak you'd ever need.
Alts are much more complicated - and FORGET any QT
crap !

I'm pretty sure I have the O'Reily 'Learning Perl/Tk' around here
someplace but I haven't used Perl in ages and see no reason to return to
it.

Next Python GUI I need to do - I'll still use Tk.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/2/25 14:37, vallor wrote:

At 2 Oct 2025 07:26:59 GMT, rbowman <bowman@montana.com> said:

On Thu, 02 Oct 2025 06:20:50 +0000, vallor wrote:

I've learned 10 times more about Tk than I ever knew before.

That's about 100 times more than I ever wanted to know. I've messed
around with Tkinter a couple of times but greatly prefer PySide6.

I'm pretty sure I have the O'Reily 'Learning Perl/Tk' around here
someplace but I haven't used Perl in ages and see no reason to return
to it.

Since the 90's, we (my company) have used perl on our backends.

Perl ... AAAAAAUUUGGGGGHH !!!

Won't touch it.

Camel dung.

With a
few caveats...for example, we have a special exception-handling class
that Carp()'s if it goes out of scope without being checked-in with.
And I think all the arguments (and hazards) for what Microsoft calls
"managed code" apply to perl.

The new fashion is to bag on perl because (reasons). One
of those reasons is inscrutable code...which is more a testament
to the programmer's ability than the language itself. Perl
can be written to be less obtuse -- see "C" for another
example of a language that can be made needlessly opaque.

Anyway, because of several decades of experience writing
perl, it's a good fit for my own personal project. I may
translate it into C++/Gtk later, I don't know.

As far as Tk goes, I picked that because it integrates well
with perl, and everything needed is in the Linux Mint repositories.
(Well, except one module, which is pure perl and readily available
on cpan.)

Also, Mrs. vallor has expressed interest in the project,
and in Usenet in general. She has a new Linux workstation
in her office, but her daily driver is (still?) a Mac Studio.
MacOS has both Tk and perl. So once I've got this thing put
together, I'm going to try to get it running on MacOS.
(Worse comes to worse, she can use XQuartz and run it
off a Linux workstation, either hers or mine.) If the
project fizzles, she may end up running pan. We'll see.

But the short answer is: perl has always been my go-to
language, but I use C when I need high-performance.

And finally: say what you will about him, I think Larry Wall is
a genius.

Look, Tk is Just Fine and has all the tweaks
you'll ever need. Also VERY well documented.

My next Python GUI app - it'll be Tk.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Fri, 3 Oct 2025 00:25:59 -0400, c186282 wrote:

Well ... I've writ a number of Tkinter Python apps.
While not 'sophisticated', Tk is "well documented" and has pretty
much every tweak you'd ever need. Alts are much more complicated -
and FORGET any QT crap !

Since my preference is for KDE, Qt fits in nicely. I also have Lubuntu on
one laptop and that uses LXQt.

The biggest problem with Qt has been the various licensing problems, hence PySide6 rather than PyQt5.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Fri, 3 Oct 2025 00:29:32 -0400, c186282 wrote:

Perl ... AAAAAAUUUGGGGGHH !!!

Won't touch it.

Camel dung.

https://www.tiobe.com/tiobe-index/

I don't remember a tiobe index for the month starting with WTF? I don't
know if I buy the theory that someone finally drove a stake through Raku's heart. Only Hurd has been a bad joke longer than Perl 6.

At this point I can't remember why I switched to Python. Maybe because
Esri started using it as a scripting language? I did like the DBI concept
and it doesn't have some of the annoying Python formatting quirks. However
it had a bushel of its own quirks. I only have a passing familiarity with
sed, awk, and some of the other stuff Wall rolled into the language.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/3/25 00:44, rbowman wrote:

On Fri, 3 Oct 2025 00:25:59 -0400, c186282 wrote:

Well ... I've writ a number of Tkinter Python apps.
While not 'sophisticated', Tk is "well documented" and has pretty
much every tweak you'd ever need. Alts are much more complicated -
and FORGET any QT crap !

Since my preference is for KDE, Qt fits in nicely. I also have Lubuntu on
one laptop and that uses LXQt.

After many frustrations/disappointments I have
totally banished QT.

The biggest problem with Qt has been the various licensing problems, hence PySide6 rather than PyQt5.

Um ... FUNCTIONAL/IMPLEMENTATION problems are my gripes.
Don't give a shit about "licensing". Will NOT use QT.

The 'better' they make things the more COMPLICATED
they make things. Not so good.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/3/25 01:06, rbowman wrote:

On Fri, 3 Oct 2025 00:29:32 -0400, c186282 wrote:

Perl ... AAAAAAUUUGGGGGHH !!!

Won't touch it.

Camel dung.

https://www.tiobe.com/tiobe-index/

I don't remember a tiobe index for the month starting with WTF? I don't
know if I buy the theory that someone finally drove a stake through Raku's heart. Only Hurd has been a bad joke longer than Perl 6.

At this point I can't remember why I switched to Python. Maybe because
Esri started using it as a scripting language? I did like the DBI concept
and it doesn't have some of the annoying Python formatting quirks. However
it had a bushel of its own quirks. I only have a passing familiarity with sed, awk, and some of the other stuff Wall rolled into the language.

Python is now VERY good ... can do almost anything
you can do with 'C' - but with more comprehensible
syntax.

So, don't cuss Python.

Anything I write, I always proto in Python.

Then, maybe, I re-write in 'C' or Pascal.
If it's not "too" I might re-do in FORTRAN
just for fun. "Speed" is the main reason for
re-do in compiled langs - not 'function'.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

ram@zedat.fu-berlin.de (Stefan Ram) writes:

Then you'd still run into the issue of figuring out how to tell the
supervisor chatbot to actually line up with what you want. So in a
way, you might just be kicking the can down the road.

That said, a single chatbot still has a ton of headroom. For my
examples I used a generic one, not something trained on a particular
programming language, coding approach, or runtime setup.

I think thatrCOs the wrong way of looking at it. They arenrCOt trained on a language as such; they are trained on a corpus of programs people wrote
in the language (and probably violating copyright in the process).
ThatrCOs why you see foolishness like using scanf to read a string: itrCOs a strategy thatrCOs well-represented in its training set.
--
https://www.greenend.org.uk/rjk/
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 03/10/2025 08:54, Richard Kettlewell wrote:

ram@zedat.fu-berlin.de (Stefan Ram) writes:

Then you'd still run into the issue of figuring out how to tell the
supervisor chatbot to actually line up with what you want. So in a
way, you might just be kicking the can down the road.

That said, a single chatbot still has a ton of headroom. For my
examples I used a generic one, not something trained on a particular
programming language, coding approach, or runtime setup.

I think thatrCOs the wrong way of looking at it. They arenrCOt trained on a language as such; they are trained on a corpus of programs people wrote
in the language (and probably violating copyright in the process).
ThatrCOs why you see foolishness like using scanf to read a string: itrCOs a strategy thatrCOs well-represented in its training set.

It does seem to me that AI in general perfectly captures the idiocy, misinformation and general incompetence of the 'average person on the internet' . It epitomises 'we all say it, so it must be true' logic.
--
Ideas are more powerful than guns. We would not let our enemies have
guns, why should we let them have ideas?

Josef Stalin

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

vallor <vallor@vallor.earth> wrote or quoted:

Call me quirky, but I wanted to write a newsreader
in Tk/perl. Off and on, over the last 6 days, I've been writing
it using ChatGPT5.
Next feature I'm going to add is threading -- but not tonight. ;)

I started to write a newsreader in Python, but then decided to
focus on implementing my markup language for some time . . .

Yesterday I noticed that my chatbot seems to be less capable in
writing Perl code than in writing Python code.

I wanted to answer a question in a programming newsgroup, where
someone wanted to provided a pattern to fill in the results of
a regexp search, like "$1 - $2", where "$1" means, "insert the
result of the first match group here".

Turned out this was hard for my chatbot who at the same time was
able to immediately provide the correct solution in Python. The
Perl version took several cycles of requests to generalize from
two matcher groups to any number of groups and debugging from my
side. Finally, I arrived at,

use strict;
use warnings;

my $quellname = '123 45';
my $muster = '(\d+)\s(\d+)';
my $ziel = '$1 - $2';

if (my @matches = $quellname =~ $muster)
{ my $zielname = $ziel;
$zielname =~ s/\$(\d+)/ do
{ my $index = $1 - 1;
$index >= 0 && $index <= $#matches ? $matches[$index] : '' } /eg;
print "rename '$quellname' -> '$zielname'\n"; }

. Well, in a sense, this post here /is/ being sent by my own
newsreader. But it's not a complete newsreader, it's just the
part for /sending/ a post. I write the post together with the
headers using a text editor (which text editor is /not/ written
by me) and save it to file. The I invoke my custom Python
script to check it, do some modifications and post it.

For example, above I still see, "vallor <vallor@vallor.earth>
writes:" with "writes", but I think my script will modify this
a bit (depending on the contents of the "Content-Language:" header
line), and what you now see above is the result. It would also
refuse to send the post if that header line were missing or be
incomprehensible to it. Here I write "I got to go" with "g-o-t-t-a",
but it's possible my script also modified this spelling.


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

vallor <vallor@vallor.earth> writes:

And finally: say what you will about him, I think Larry Wall is
a genius.

Hear, Hear!
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Fri, 3 Oct 2025 10:02:28 +0100, The Natural Philosopher wrote:

It does seem to me that AI in general perfectly captures the idiocy, misinformation and general incompetence of the 'average person on the internet' . It epitomises 'we all say it, so it must be true' logic.

Do others say that besides you?
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 3 Oct 2025 11:38:42 GMT, Stefan Ram wrote:

Yesterday I noticed that my chatbot seems to be less capable in
writing Perl code than in writing Python code.

Did you try PHP?
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Fri, 3 Oct 2025 11:33:45 -0000 (UTC), Bob Vloon wrote:

vallor <vallor@vallor.earth> writes:

And finally: say what you will about him, I think Larry Wall is a
genius.

Hear, Hear!

He might be remembered more for creating patch(1) than Perl. There had
been tools to apply file edits before, but the clever thing about
patch was that it could merge patches into files that had already had
patches applied from other sources.

As a result, the concept of distributing and applying source-code
diffs has become the basic currency of collaborative open-source
development.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

Lawrence =?iso-8859-13?q?D=FFOliveiro?= <ldo@nz.invalid> wrote or quoted:

On 3 Oct 2025 11:38:42 GMT, Stefan Ram wrote:

Yesterday I noticed that my chatbot seems to be less capable in
writing Perl code than in writing Python code.

Did you try PHP?

Right now I have no use for PHP. But if I remember right, there's
this guy hanging around on the German Usenet who says he doesn't
know PHP and he's fine with the PHP the chatbot writes for him.


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 3 Oct 2025 20:59:44 GMT, Stefan Ram wrote:

On Fri, 3 Oct 2025 20:42:17 -0000 (UTC), Lawrence DrCOOliveiro wrote:

On 3 Oct 2025 11:38:42 GMT, Stefan Ram wrote:

Yesterday I noticed that my chatbot seems to be less capable in
writing Perl code than in writing Python code.

Did you try PHP?

Right now I have no use for PHP.

I was asking if the chatbot was better versed in PHP than it was in
Perl.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/3/25 03:54, Richard Kettlewell wrote:

ram@zedat.fu-berlin.de (Stefan Ram) writes:

Then you'd still run into the issue of figuring out how to tell the
supervisor chatbot to actually line up with what you want. So in a
way, you might just be kicking the can down the road.

That said, a single chatbot still has a ton of headroom. For my
examples I used a generic one, not something trained on a particular
programming language, coding approach, or runtime setup.

I think thatrCOs the wrong way of looking at it. They arenrCOt trained on a language as such; they are trained on a corpus of programs people wrote
in the language (and probably violating copyright in the process).
ThatrCOs why you see foolishness like using scanf to read a string: itrCOs a strategy thatrCOs well-represented in its training set.

Um, yea, NEVER use 'scanf' these days !

As said somewhere, the weekly M$ security warnings
are just full of 'buffer overflow' warnings. They
won't, or can't, fix it all.

There are input-length-limited functions in 'C' and
a lot of other langs ... nobody uses them.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/3/25 05:02, The Natural Philosopher wrote:

On 03/10/2025 08:54, Richard Kettlewell wrote:

ram@zedat.fu-berlin.de (Stefan Ram) writes:

-a-a Then you'd still run into the issue of figuring out how to tell the >>> -a-a supervisor chatbot to actually line up with what you want.-a So in a >>> -a-a way, you might just be kicking the can down the road.

-a-a That said, a single chatbot still has a ton of headroom. For my
-a-a examples I used a generic one, not something trained on a particular >>> -a-a programming language, coding approach, or runtime setup.

I think thatrCOs the wrong way of looking at it. They arenrCOt trained on a >> language as such; they are trained on a corpus of programs people wrote
in the language (and probably violating copyright in the process).
ThatrCOs why you see foolishness like using scanf to read a string: itrCOs a >> strategy thatrCOs well-represented in its training set.

It does seem to me that AI in general perfectly captures the idiocy, misinformation and general incompetence of the 'average person on the internet' . It epitomises 'we all say it, so it must be true' logic.

Well, today's "AI" DOES learn from what's Out There -
including all the serious mistakes.


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/3/25 16:41, Lawrence DrCOOliveiro wrote:

On Fri, 3 Oct 2025 10:02:28 +0100, The Natural Philosopher wrote:

It does seem to me that AI in general perfectly captures the idiocy,
misinformation and general incompetence of the 'average person on the
internet' . It epitomises 'we all say it, so it must be true' logic.

Do others say that besides you?

I'll vote in favor.

"AI" is trained on vast quantities of 'human stuff',
INCLUDING all the errors.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/3/25 16:42, Lawrence DrCOOliveiro wrote:

On 3 Oct 2025 11:38:42 GMT, Stefan Ram wrote:

Yesterday I noticed that my chatbot seems to be less capable in
writing Perl code than in writing Python code.

Did you try PHP?

PHP is annoying. Had to write several web sites
that wound up being 75% PHP just to add enough
IQ to be worth it.

Perl is camel dung.

Python is a jack of all trades at this point
and the net is packed full of examples - good,
bad and otherwise.

"AI" is trained by pirating all that code.

I wonder if AI knows to use the 2nd/3rd problem
solutions - which are usually better/tighter ?
Probably not.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/3/25 16:59, Stefan Ram wrote:

Lawrence =?iso-8859-13?q?D=FFOliveiro?= <ldo@nz.invalid> wrote or quoted:

On 3 Oct 2025 11:38:42 GMT, Stefan Ram wrote:

Yesterday I noticed that my chatbot seems to be less capable in
writing Perl code than in writing Python code.

Did you try PHP?

Right now I have no use for PHP. But if I remember right, there's
this guy hanging around on the German Usenet who says he doesn't
know PHP and he's fine with the PHP the chatbot writes for him.

Uh oh ........

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-04, c186282 <c186282@nnada.net> wrote:

Um, yea, NEVER use 'scanf' these days !

As said somewhere, the weekly M$ security warnings
are just full of 'buffer overflow' warnings. They
won't, or can't, fix it all.

There are input-length-limited functions in 'C' and
a lot of other langs ... nobody uses them.

Speak for yourself. I've been reading files with fgets()
for decades, parsing them myself. Over the last few years,
though, I've replaced fgets() with getline(), which lifts
length restrictions entirely.
--
/~\ Charlie Gibbs | Growth for the sake of
\ / <cgibbs@kltpzyxm.invalid> | growth is the ideology
X I'm really at ac.dekanfrus | of the cancer cell.
/ \ if you read it the right way. | -- Edward Abbey
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

Charlie Gibbs <cgibbs@kltpzyxm.invalid> wrote at 03:53 this Wednesday (GMT):

On 2025-10-01, Stefan Ram <ram@zedat.fu-berlin.de> wrote:

Many error checks are added to make error situations managable to
non-programming users of a program. When a programmer writes code
for his own use, he can sometimes tolerate more cryptic raw error
messages as they come from the runtime or consciously decide to
take some risks when he himself would be their only victim.

You mean more cyptic than "Something went wrong"?

It used to be that a singularly unhelpful message like
"It didn't work" only came from Users from Hell.
Nowadays I'm seeing it in more and more software -
and that goes in spades for web pages.

I half believe they do that to keep people from fixing things themselves
and instead go and contact a support agent (that costs money).
--
user <candycane> is generated from /dev/urandom
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/10/25 14:00, candycanearter07 wrote:

Charlie Gibbs <cgibbs@kltpzyxm.invalid> wrote at 03:53 this Wednesday (GMT):

On 2025-10-01, Stefan Ram <ram@zedat.fu-berlin.de> wrote:

Many error checks are added to make error situations managable to
non-programming users of a program. When a programmer writes code
for his own use, he can sometimes tolerate more cryptic raw error
messages as they come from the runtime or consciously decide to
take some risks when he himself would be their only victim.

You mean more cyptic than "Something went wrong"?

It used to be that a singularly unhelpful message like
"It didn't work" only came from Users from Hell.
Nowadays I'm seeing it in more and more software -
and that goes in spades for web pages.

I half believe they do that to keep people from fixing things themselves
and instead go and contact a support agent (that costs money).

"When in question, when in doubt, follow
the MONEY and you'll find out"'

- Me, like 20+ years ago

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 11/10/2025 07:41, c186282 wrote:

"When in question, when in doubt, follow
-a the MONEY and you'll find out"'

Ah. a fellow cynic.
--
rCLThe fundamental cause of the trouble in the modern world today is that
the stupid are cocksure while the intelligent are full of doubt."

- Bertrand Russell


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Sat, 11 Oct 2025 02:41:10 -0400, c186282 wrote:

"When in question, when in doubt, follow the MONEY and you'll find
out"'

The Franklins do leave a discernible trail.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Sat, 11 Oct 2025 11:00:04 +0100, The Natural Philosopher wrote:

On 11/10/2025 07:41, c186282 wrote:

"When in question, when in doubt, follow the MONEY and you'll find
out"'

Ah. a fellow cynic.

rCLFollow the moneyrCY ... rCLCui bono?rCY ... great guides, I find, to deconstructing popular conspiracy theories in particular.

But the conspiracy theorists donrCOt like it when I point out how many of their pet beliefs donrCOt add up when analyzed this way.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/11/25 15:03, rbowman wrote:

On Sat, 11 Oct 2025 02:41:10 -0400, c186282 wrote:

"When in question, when in doubt, follow the MONEY and you'll find
out"'

The Franklins do leave a discernible trail.

Came up with that one myself ... like 30+ years ago.
Irresistible rhyme :-)

Actually, think I came up with it on the CompuServe
Forums ........

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/11/25 15:39, Lawrence DrCOOliveiro wrote:

On Sat, 11 Oct 2025 11:00:04 +0100, The Natural Philosopher wrote:

On 11/10/2025 07:41, c186282 wrote:

"When in question, when in doubt, follow the MONEY and you'll find
out"'

Ah. a fellow cynic.

rCLFollow the moneyrCY ... rCLCui bono?rCY ... great guides, I find, to deconstructing popular conspiracy theories in particular.

But the conspiracy theorists donrCOt like it when I point out how many of their pet beliefs donrCOt add up when analyzed this way.

THIS one DOES.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 12/10/2025 03:28, c186282 wrote:

On 10/11/25 15:03, rbowman wrote:

On Sat, 11 Oct 2025 02:41:10 -0400, c186282 wrote:

-a-a-a "When in question, when in doubt, follow the MONEY and you'll find >>> -a-a-a out"'

The Franklins do leave a discernible trail.

-a Came up with that one myself ... like 30+ years ago.
-a Irresistible rhyme-a :-)

-a Actually, think I came up with it on the CompuServe
-a Forums ........

When as a student we could only afford to repair our own cars, I came up
with similar
'when in doubt, rip it out'

The time spent removing an entire axle to get a brake drum off and onto
the bench with total access and proper tools was way less that lying
upside down with a hammer on the floor getting rust flakes in your eye....
--
The urge to save humanity is almost always a false front for the urge to
rule.
rCo H. L. Mencken, American journalist, 1880-1956

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 12/10/2025 03:31, c186282 wrote:

On 10/11/25 15:39, Lawrence DrCOOliveiro wrote:

On Sat, 11 Oct 2025 11:00:04 +0100, The Natural Philosopher wrote:

On 11/10/2025 07:41, c186282 wrote:

"When in question, when in doubt, follow the MONEY and you'll find
out"'

Ah. a fellow cynic.

rCLFollow the moneyrCY ... rCLCui bono?rCY ... great guides, I find, to
deconstructing popular conspiracy theories in particular.

But the conspiracy theorists donrCOt like it when I point out how many of
their pet beliefs donrCOt add up when analyzed this way.

-a THIS one DOES.

Which is why its not a 'conspiracy theory' in the normal sense.

I have met with and worked with people whose duplicity, and
collaboration to deceive their customers, the stock market and indeed
the Law, is simply astonishing to a normal human being brought up in a
broadly moral sort of way.

Some like money, others like power. Some want political power. Others
want fame.
The overriding characteristic is that they all want more than can be got
by honesty hard work and intelligence. So they cheat.
--
There is nothing a fleet of dispatchable nuclear power plants cannot do
that cannot be done worse and more expensively and with higher carbon emissions and more adverse environmental impact by adding intermittent renewable energy.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Sun, 12 Oct 2025 12:04:42 +0100, The Natural Philosopher wrote:

On 12/10/2025 03:28, c186282 wrote:

On 10/11/25 15:03, rbowman wrote:

On Sat, 11 Oct 2025 02:41:10 -0400, c186282 wrote:

-a-a-a "When in question, when in doubt, follow the MONEY and you'll
-a-a-a find out"'

The Franklins do leave a discernible trail.

-a Came up with that one myself ... like 30+ years ago.
-a Irresistible rhyme-a :-)

-a Actually, think I came up with it on the CompuServe Forums ........

When as a student we could only afford to repair our own cars, I came up
with similar 'when in doubt, rip it out'

The time spent removing an entire axle to get a brake drum off and onto
the bench with total access and proper tools was way less that lying
upside down with a hammer on the floor getting rust flakes in your
eye....

That could escalate. I had a Plymouth with a dying automatic transmission, that I decided to replace with a manual. That required some creative
design work to get a clutch pedal along with a hydraulic actuator. Floor
shift kits were available so a little hacking through the floor boards and building sort of a console out of sheet metal took care of that. Then
there was shortening the drive shaft.

Finally the project was done and all was well or at least until a cop
stopped me for a tail light problem. He insisted that I demonstrate the emergency brake's operation. The AT had a drum on the tailshaft that was
the emergency brake but that was long gone. Next step was acquiring a rear axle that had the necessary brake mechanism, and fabricating an activation cable.

I had a lot more time than money. It also helped that the friendly family physician was freely prescribing 'diet pills'.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/12/25 07:04, The Natural Philosopher wrote:

On 12/10/2025 03:28, c186282 wrote:

On 10/11/25 15:03, rbowman wrote:

On Sat, 11 Oct 2025 02:41:10 -0400, c186282 wrote:

-a-a-a "When in question, when in doubt, follow the MONEY and you'll find >>>> -a-a-a out"'

The Franklins do leave a discernible trail.

-a-a Came up with that one myself ... like 30+ years ago.
-a-a Irresistible rhyme-a :-)

-a-a Actually, think I came up with it on the CompuServe
-a-a Forums ........

When as a student we could only afford to repair our own cars, I came up with similar
'when in doubt, rip it out'

The time spent removing an entire axle to get a brake drum off and onto
the bench with total access and proper tools was way less that lying
upside down with a hammer on the floor getting rust flakes in your eye....

Not every designer/corp thinks in terms
of maint.

Most USA cars, you can easily get to, replace,
things like brake drums/discs/bearings without
getting more extreme. But not EVERYBODY thought
that way.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 13/10/2025 08:24, c186282 wrote:

On 10/12/25 07:04, The Natural Philosopher wrote:

On 12/10/2025 03:28, c186282 wrote:

On 10/11/25 15:03, rbowman wrote:

On Sat, 11 Oct 2025 02:41:10 -0400, c186282 wrote:

-a-a-a "When in question, when in doubt, follow the MONEY and you'll find >>>>> -a-a-a out"'

The Franklins do leave a discernible trail.

-a-a Came up with that one myself ... like 30+ years ago.
-a-a Irresistible rhyme-a :-)

-a-a Actually, think I came up with it on the CompuServe
-a-a Forums ........

When as a student we could only afford to repair our own cars, I came
up with similar
'when in doubt, rip it out'

The time spent removing an entire axle to get a brake drum off and
onto the bench with total access and proper tools was way less that
lying upside down with a hammer on the floor getting rust flakes in
your eye....

-a Not every designer/corp thinks in terms
-a of maint.

-a Most USA cars, you can easily get to, replace,
-a things like brake drums/discs/bearings without
-a getting more extreme. But not EVERYBODY thought
-a that way.

Stuff rusts up no matter how easy it is to get to...
--
Canada is all right really, though not for the whole weekend.

"Saki"

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

Most USA cars, you can easily get to, replace, things like brake
drums/discs/bearings without getting more extreme. But not EVERYBODY
thought that way.

Yeah, sure... When the master cylinder on my '51 Chevy failed it became obvious that the cylinder had been bolted to the frame before the body was added. There was a small plate to allow replacing fluid but no provision
for R&R.

I'm one of those strange people who looks under the hood when buying a
car. I like to identify everything that may need maintenance and make sure it's accessible.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

Most USA cars, you can easily get to, replace, things like brake
drums/discs/bearings without getting more extreme. But not EVERYBODY
thought that way.

Yeah, sure... When the master cylinder on my '51 Chevy failed it became obvious that the cylinder had been bolted to the frame before the body was added. There was a small plate to allow replacing fluid but no provision
for R&R.

Hey, 1951 ....

Drill the bolts and proceed.

I'm one of those strange people who looks under the hood when buying a
car. I like to identify everything that may need maintenance and make sure it's accessible.

THESE days ??? They INTENTIONALLY hide everything.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Tue, 14 Oct 2025 01:24:59 -0400, c186282 wrote:

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

Most USA cars, you can easily get to, replace, things like brake
drums/discs/bearings without getting more extreme. But not
EVERYBODY thought that way.

Yeah, sure... When the master cylinder on my '51 Chevy failed it
became obvious that the cylinder had been bolted to the frame before
the body was added. There was a small plate to allow replacing fluid
but no provision for R&R.

Hey, 1951 ....

Drill the bolts and proceed.

Yeah, a '51 introduced me to interesting technology like kingpin reamers.
otoh I paid $35 for it and had a lot of fun in the car. Can't beat that
old straight 6.

My family had a new '51 when we drove across country. In her version of a pilot's automatic rough she would start worrying about the tappets in the middle of Wyoming. I'm not sure she even knew what a tappet was. Before hydraulic lifters a noisy valve train was a happy valve train but with her wolf's ear every click portended disaster.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-14 07:24, c186282 wrote:

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

-a-a-a Most USA cars, you can easily get to, replace, things like brake
-a-a-a drums/discs/bearings without getting more extreme. But not EVERYBODY >>> -a-a-a thought that way.

Yeah, sure...-a-a When the master cylinder on my '51 Chevy failed it became >> obvious that the cylinder had been bolted to the frame before the body
was
added. There was a small plate to allow replacing fluid but no provision
for R&R.

-a Hey, 1951 ....

-a Drill the bolts and proceed.

I'm one of those strange people who looks under the hood when buying a
car. I like to identify everything that may need maintenance and make
sure
it's accessible.

-a THESE days ??? They INTENTIONALLY hide everything.

These days what breaks are the electronic sensors. Mechanical is
eternal, almost :-)
--
Cheers, Carlos.
ESEfc-Efc+, EUEfc-Efc|;
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 14/10/2025 09:04, Carlos E.R. wrote:

On 2025-10-14 07:24, c186282 wrote:

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

-a-a-a Most USA cars, you can easily get to, replace, things like brake >>>> -a-a-a drums/discs/bearings without getting more extreme. But not
EVERYBODY
-a-a-a thought that way.

Yeah, sure...-a-a When the master cylinder on my '51 Chevy failed it
became
obvious that the cylinder had been bolted to the frame before the
body was
added. There was a small plate to allow replacing fluid but no provision >>> for R&R.

-a-a Hey, 1951 ....

-a-a Drill the bolts and proceed.

I'm one of those strange people who looks under the hood when buying a
car. I like to identify everything that may need maintenance and make
sure
it's accessible.

-a-a THESE days ??? They INTENTIONALLY hide everything.

These days what breaks are the electronic sensors. Mechanical is
eternal, almost :-)

Not if its made of *plastic*.

Inlet manifolds made of plastic at a $600 replacement cost? Who, ever,
broke a cast inlet manifold back in the day?...

Composite timing belts that snap every 100,000 miles...

Engines that destroy themselves in a manufacturer's attempt to get
better gas mileage or emissions?

The cars are designed to last the warranty period and then be too
expensive to be worth fixing.
--
In a Time of Universal Deceit, Telling the Truth Is a Revolutionary Act.

- George Orwell

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-14 11:04, The Natural Philosopher wrote:

On 14/10/2025 09:04, Carlos E.R. wrote:

On 2025-10-14 07:24, c186282 wrote:

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

-a-a-a Most USA cars, you can easily get to, replace, things like brake >>>>> -a-a-a drums/discs/bearings without getting more extreme. But not
EVERYBODY
-a-a-a thought that way.

Yeah, sure...-a-a When the master cylinder on my '51 Chevy failed it
became
obvious that the cylinder had been bolted to the frame before the
body was
added. There was a small plate to allow replacing fluid but no
provision
for R&R.

-a-a Hey, 1951 ....

-a-a Drill the bolts and proceed.

I'm one of those strange people who looks under the hood when buying a >>>> car. I like to identify everything that may need maintenance and
make sure
it's accessible.

-a-a THESE days ??? They INTENTIONALLY hide everything.

These days what breaks are the electronic sensors. Mechanical is
eternal, almost :-)

Not if its made of *plastic*.

Hum. Yes. I repaired a radiator fan attachment part to the car body,
with epoxy resin with a fibre glass filling. The mechanic said *he*
could not do that (because he could not guarantee the repair for the
mandatory time), but me, sure, fine. The repair did hold till the car
was recycled.

Inlet manifolds made of plastic at a $600 replacement cost? Who, ever,
broke a cast inlet manifold back in the day?...

Composite timing belts that snap every 100,000 miles...

Some are submerged in oil and break earlier. They decompose, clog the
oil filter with debris, and the engine is totalled because of oil
failure. Who was the idiot that thought submerging a rubber belt in oil
was a good idea?

Engines that destroy themselves in a manufacturer's attempt to get
better gas mileage or emissions?

The cars are designed to last the warranty period and then be too
expensive to be worth fixing.

--
Cheers, Carlos.
ESEfc-Efc+, EUEfc-Efc|;
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 14/10/2025 11:06, Carlos E.R. wrote:

On 2025-10-14 11:04, The Natural Philosopher wrote:

On 14/10/2025 09:04, Carlos E.R. wrote:

On 2025-10-14 07:24, c186282 wrote:

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

-a-a-a Most USA cars, you can easily get to, replace, things like brake >>>>>> -a-a-a drums/discs/bearings without getting more extreme. But not >>>>>> EVERYBODY
-a-a-a thought that way.

Yeah, sure...-a-a When the master cylinder on my '51 Chevy failed it >>>>> became
obvious that the cylinder had been bolted to the frame before the
body was
added. There was a small plate to allow replacing fluid but no
provision
for R&R.

-a-a Hey, 1951 ....

-a-a Drill the bolts and proceed.

I'm one of those strange people who looks under the hood when buying a >>>>> car. I like to identify everything that may need maintenance and
make sure
it's accessible.

-a-a THESE days ??? They INTENTIONALLY hide everything.

These days what breaks are the electronic sensors. Mechanical is
eternal, almost :-)

Not if its made of *plastic*.

Hum. Yes. I repaired a radiator fan attachment part to the car body,
with epoxy resin with a fibre glass filling. The mechanic said *he*
could not do that (because he could not guarantee the repair for the mandatory time), but me, sure, fine. The repair did hold till the car
was recycled.

I remember my first encounter with a Ford radiator drain plug made on nylon. The aftermarket brass one was *cheaper* than the replacement Ford part...

Inlet manifolds made of plastic at a $600 replacement cost? Who, ever,
broke a cast inlet manifold back in the day?...

Composite timing belts that snap every 100,000 miles...

Some are submerged in oil and break earlier. They decompose, clog the
oil filter with debris, and the engine is totalled because of oil
failure. Who was the idiot that thought submerging a rubber belt in oil
was a good idea?

Well you are supposed to *change* them, and the oil they run in,
regularly...

You should watch the You tube channel 'I do cars' to see why regular oil changes and maintenance are a good idea.
The USA, having at best sketchy vehicle mandatory inspections, is a rich source of 'what happens if I drive it till it falls apart?' vehicles.

I used to manage a very small fleet of company cars. We would lease them
and throw them back after a couple of years. Now mine did get
maintained, but many people in that game would not. Beyond necessity.
Tyres and brakes and that's it.

Beware the two year old car with 120,000 miles on the clock that has
*never had an oil change*
--
"Nature does not give up the winter because people dislike the cold."

rCo Confucius

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-14 12:32, The Natural Philosopher wrote:

On 14/10/2025 11:06, Carlos E.R. wrote:

Composite timing belts that snap every 100,000 miles...

Some are submerged in oil and break earlier. They decompose, clog the
oil filter with debris, and the engine is totalled because of oil
failure. Who was the idiot that thought submerging a rubber belt in
oil was a good idea?

Well you are supposed to *change* them, and the oil they run in, regularly...

Even if you do, they break way before their due time.

It is a known issue, well published.
--
Cheers, Carlos.
ESEfc-Efc+, EUEfc-Efc|;
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Tue, 14 Oct 2025 10:04:57 +0100, The Natural Philosopher wrote:

Composite timing belts that snap every 100,000 miles...

Like timing chains never failed... One difference was there were few interference engines.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Tue, 14 Oct 2025 11:32:06 +0100, The Natural Philosopher wrote:

The USA, having at best sketchy vehicle mandatory inspections, is a rich source of 'what happens if I drive it till it falls apart?' vehicles.

Like many things, this is on a state by state basis. This state has no inspection and there are surprisingly few rolling wrecks. New Hampshire
had semi-annual inspections that could be a pain in the ass. I bought an
old Dodge pickup and somewhere along the line someone with big truck aspirations had added clearance lights to the top of the cab. Completely nonstandard and not legally required but because they were there they had
to work. I've also been flagged for lack of a horn. The only time I use a
horn is by mistake but you had to have one.

They were good business for repair shops. The dreaded 'your ball joints
are worn' was good for several hundred dollars.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Tue, 14 Oct 2025 12:45:07 +0200, Carlos E.R. wrote:

On 2025-10-14 12:32, The Natural Philosopher wrote:

On 14/10/2025 11:06, Carlos E.R. wrote:

Composite timing belts that snap every 100,000 miles...

Some are submerged in oil and break earlier. They decompose, clog the
oil filter with debris, and the engine is totalled because of oil
failure. Who was the idiot that thought submerging a rubber belt in
oil was a good idea?

Well you are supposed to *change* them, and the oil they run in,
regularly...

Even if you do, they break way before their due time.

It is a known issue, well published.

So far that seems to be mostly a European thing. The engine in my Toyota
uses a chain. I did have a Geo, which was a rebadged Suzuki, had a dry
belt. I didn't know the car's history but it was getting up toward 100K
miles so I changed it. It was tight quarters but I've had worse jobs.

The belt on my Harley failed at 47K. It was quite odd. I came out from
work, jumped on the bike, put it in gear, and nothing happened. The belt
broke straight across with no real sign of damage. I'd always worried
about rocks when riding on gravel roads but didn't see any evidence of
that.

The new belt was about $150. That sounds pricey but compared to replacing chains and sometimes the sprocket plus the maintenance it was a good deal
in the long run. Like belt in oil systems it was driven by the need to
meet the noise requirements rather than a pure engineering decision.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-14, rbowman <bowman@montana.com> wrote:

On Tue, 14 Oct 2025 11:32:06 +0100, The Natural Philosopher wrote:

The USA, having at best sketchy vehicle mandatory inspections, is a rich
source of 'what happens if I drive it till it falls apart?' vehicles.

Like many things, this is on a state by state basis. This state has no inspection and there are surprisingly few rolling wrecks. New Hampshire
had semi-annual inspections that could be a pain in the ass. I bought an
old Dodge pickup and somewhere along the line someone with big truck aspirations had added clearance lights to the top of the cab. Completely nonstandard and not legally required but because they were there they had
to work. I've also been flagged for lack of a horn. The only time I use a horn is by mistake but you had to have one.

They were good business for repair shops. The dreaded 'your ball joints
are worn' was good for several hundred dollars.

"They've got you by the ball joints."

We used to have vehicle inspection stations here in B.C. but the
provincial government shut them down around 1982 as a "cost-saving
measure". (Compared to the billions they're tossing around these
days, it seems minor now.) They did things like check brakes for
adequate and straight braking, check the exhaust system for leaks,
check the windshield for cracks, etc. The inspection was mandatory
once a year, and you got a decal in the corner of your windshield
to prove to passing policemen that you passed.

It was pretty much a given that you'd flunk on headlights if
nothing else. There was a service station next door to each
inspection station that did a good business adjusting headlights -
they had the same equipment that the inspection stations used.

I wish the inspection stations were still around. Most modern
cars would fail due to overly-bright headlights aimed too high.
(Gotta dazzle the oncoming traffic, doncha know - the car ads
say it's cool.) When we cross the border into Washington I
notice that headlights down there aren't nearly as obnoxious.
In fact, the last time I bought a replacement headlamp I noticed
that a number of them were marked "Legal in Canada and Mexico
only". Apparently this is something the States got right.
--
/~\ Charlie Gibbs | Growth for the sake of
\ / <cgibbs@kltpzyxm.invalid> | growth is the ideology
X I'm really at ac.dekanfrus | of the cancer cell.
/ \ if you read it the right way. | -- Edward Abbey
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-14, rbowman <bowman@montana.com> wrote:

<timing belts>

I once had a 1984 Civic whose water pump seized. It stripped
all the teeth off the timing belt - except for about a quarter
of an inch along one edge which was just enough to keep the
camshaft turning. I was able to limp to a mechanic and get
it replaced, rather than trashing the engine. I tip my hat
to the Honda engineers for that one - I like to think that
they deliberately offset the pulleys a bit for just that reason.
--
/~\ Charlie Gibbs | Growth for the sake of
\ / <cgibbs@kltpzyxm.invalid> | growth is the ideology
X I'm really at ac.dekanfrus | of the cancer cell.
/ \ if you read it the right way. | -- Edward Abbey
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/14/25 03:09, rbowman wrote:

On Tue, 14 Oct 2025 01:24:59 -0400, c186282 wrote:

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

Most USA cars, you can easily get to, replace, things like brake
drums/discs/bearings without getting more extreme. But not
EVERYBODY thought that way.

Yeah, sure... When the master cylinder on my '51 Chevy failed it
became obvious that the cylinder had been bolted to the frame before
the body was added. There was a small plate to allow replacing fluid
but no provision for R&R.

Hey, 1951 ....

Drill the bolts and proceed.

Yeah, a '51 introduced me to interesting technology like kingpin reamers. otoh I paid $35 for it and had a lot of fun in the car. Can't beat that
old straight 6.

$35 ??? That'd be near $500 now ! Expensive tool !

The sixes weren't hot-rods ... but they were
good enough for most stuff and cheaper than
V-8s.

My family had a new '51 when we drove across country. In her version of a pilot's automatic rough she would start worrying about the tappets in the middle of Wyoming. I'm not sure she even knew what a tappet was. Before hydraulic lifters a noisy valve train was a happy valve train but with her wolf's ear every click portended disaster.

The valve adjustments tended to drift in those
old engines rather quickly. So, 'tapppety-tap'
was hardly unusual. Dangerous ? Rarely. Just
gets more and more annoying.

Was never sure WHY the adjustments didn't hold
up better ... points seemed to have a similar
issue. I was happy to see magnetic sensors
take over.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/14/25 05:04, The Natural Philosopher wrote:

On 14/10/2025 09:04, Carlos E.R. wrote:

On 2025-10-14 07:24, c186282 wrote:

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

-a-a-a Most USA cars, you can easily get to, replace, things like brake >>>>> -a-a-a drums/discs/bearings without getting more extreme. But not
EVERYBODY
-a-a-a thought that way.

Yeah, sure...-a-a When the master cylinder on my '51 Chevy failed it
became
obvious that the cylinder had been bolted to the frame before the
body was
added. There was a small plate to allow replacing fluid but no
provision
for R&R.

-a-a Hey, 1951 ....

-a-a Drill the bolts and proceed.

I'm one of those strange people who looks under the hood when buying a >>>> car. I like to identify everything that may need maintenance and
make sure
it's accessible.

-a-a THESE days ??? They INTENTIONALLY hide everything.

These days what breaks are the electronic sensors. Mechanical is
eternal, almost :-)

Not if its made of *plastic*.

Inlet manifolds made of plastic at a $600 replacement cost? Who, ever,
broke a cast inlet manifold back in the day?...

Composite timing belts that snap every 100,000 miles...

Engines that destroy themselves in a manufacturer's attempt to get
better gas mileage or emissions?

The cars are designed to last the warranty period and then be too
expensive to be worth fixing.

Kind of, yes. 'Premium' brands may do a little better.

Oh well, scan the antiques sites - if you want a 57 Nash
you can BUY one still. Parts may be a bit few and far
between now however.

Ya know, with modern 3-D metal printing, somebody ought
to sell one-off duplicates for almost every part for
old cars. All they need are the original dimensions.
Heat-treated parts, not that many in old vehicles,
would be more of a challenge.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/14/25 06:06, Carlos E.R. wrote:

On 2025-10-14 11:04, The Natural Philosopher wrote:

On 14/10/2025 09:04, Carlos E.R. wrote:

On 2025-10-14 07:24, c186282 wrote:

On 10/13/25 15:05, rbowman wrote:

On Mon, 13 Oct 2025 03:24:19 -0400, c186282 wrote:

-a-a-a Most USA cars, you can easily get to, replace, things like brake >>>>>> -a-a-a drums/discs/bearings without getting more extreme. But not >>>>>> EVERYBODY
-a-a-a thought that way.

Yeah, sure...-a-a When the master cylinder on my '51 Chevy failed it >>>>> became
obvious that the cylinder had been bolted to the frame before the
body was
added. There was a small plate to allow replacing fluid but no
provision
for R&R.

-a-a Hey, 1951 ....

-a-a Drill the bolts and proceed.

I'm one of those strange people who looks under the hood when buying a >>>>> car. I like to identify everything that may need maintenance and
make sure
it's accessible.

-a-a THESE days ??? They INTENTIONALLY hide everything.

These days what breaks are the electronic sensors. Mechanical is
eternal, almost :-)

Not if its made of *plastic*.

Hum. Yes. I repaired a radiator fan attachment part to the car body,
with epoxy resin with a fibre glass filling. The mechanic said *he*
could not do that (because he could not guarantee the repair for the mandatory time), but me, sure, fine. The repair did hold till the car
was recycled.

Inlet manifolds made of plastic at a $600 replacement cost? Who, ever,
broke a cast inlet manifold back in the day?...

Composite timing belts that snap every 100,000 miles...

Some are submerged in oil and break earlier. They decompose, clog the
oil filter with debris, and the engine is totalled because of oil
failure. Who was the idiot that thought submerging a rubber belt in oil
was a good idea?

It was "good" because it was FAST AND CHEAP.

By the mid 70s the US auto industry came under
intense pressure from Europe and especially
from Japan. This meant 'cost saving' measures
whether they were particularly great or not.
The good old 'heavy metal solutions' cost too
much and corner-cutting became the norm.

Engines that destroy themselves in a manufacturer's attempt to get
better gas mileage or emissions?

The cars are designed to last the warranty period and then be too
expensive to be worth fixing.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On Tue, 14 Oct 2025 23:31:45 GMT, Charlie Gibbs wrote:

I wish the inspection stations were still around. Most modern cars
would fail due to overly-bright headlights aimed too high. (Gotta dazzle
the oncoming traffic, doncha know - the car ads say it's cool.) When we cross the border into Washington I notice that headlights down there
aren't nearly as obnoxious. In fact, the last time I bought a
replacement headlamp I noticed that a number of them were marked "Legal
in Canada and Mexico only". Apparently this is something the States got right.

The HID systems are really annoying to say nothing of pickups with
headlights at eye level if you're in a passenger car. I had cataracts
removed from both eyes last year. Prior to that I was mostly flying blind
with oncoming headlights.

I'm not big on gadgetry and my Toyota doesn't have remote entry, lane
keeping, and all that but one thing it does have is automatic headlight dimming. In almost all cases it dims the lights when I would manually
either overtaking or approaching. One quirk is a 20 mph curve on my way
home that has a lot of reflective signage that will trigger dimming.

iirc Cadillac had something like that about 50 years ago that didn't work
very well. I assume this system has more intelligence and better sensors.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 10/15/25 02:41, rbowman wrote:

On Tue, 14 Oct 2025 23:31:45 GMT, Charlie Gibbs wrote:

I wish the inspection stations were still around. Most modern cars
would fail due to overly-bright headlights aimed too high. (Gotta dazzle
the oncoming traffic, doncha know - the car ads say it's cool.) When we
cross the border into Washington I notice that headlights down there
aren't nearly as obnoxious. In fact, the last time I bought a
replacement headlamp I noticed that a number of them were marked "Legal
in Canada and Mexico only". Apparently this is something the States got
right.

The HID systems are really annoying to say nothing of pickups with
headlights at eye level if you're in a passenger car. I had cataracts
removed from both eyes last year. Prior to that I was mostly flying blind with oncoming headlights.

I'm not big on gadgetry and my Toyota doesn't have remote entry, lane keeping, and all that but one thing it does have is automatic headlight dimming. In almost all cases it dims the lights when I would manually
either overtaking or approaching. One quirk is a 20 mph curve on my way
home that has a lot of reflective signage that will trigger dimming.

iirc Cadillac had something like that about 50 years ago that didn't work very well. I assume this system has more intelligence and better sensors.

The solution to the headlights is easy, and could
have been done 50 years ago. Crossed polarization
film - one way for headlights and the other built
into the windshield. Even a 30-50 percent glare
reduction would be very useful.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 14/10/2025 20:12, rbowman wrote:

On Tue, 14 Oct 2025 10:04:57 +0100, The Natural Philosopher wrote:

Composite timing belts that snap every 100,000 miles...

Like timing chains never failed... One difference was there were few interference engines.

Timing chains rattled like SOBs long before they failed
--
rCLProgress is precisely that which rules and regulations did not foresee,rCY

rCo Ludwig von Mises

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 15/10/2025 02:06, c186282 wrote:

Ya know, with modern 3-D metal printing, somebody ought
-a to sell one-off duplicates for almost every part for
-a old cars. All they need are the original dimensions.
-a Heat-treated parts, not that many in old vehicles,
-a would be more of a challenge.

In the UK every part is available for 1970s era MGBs and the old Austin
Healy Sprite, and you can build a brand new one.
As long as you have the paperwork to say its really 50 years old, and so
avoid emissions limits
--
The New Left are the people they warned you about.

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-15 00:05, rbowman wrote:

On Tue, 14 Oct 2025 12:45:07 +0200, Carlos E.R. wrote:

On 2025-10-14 12:32, The Natural Philosopher wrote:

On 14/10/2025 11:06, Carlos E.R. wrote:

Composite timing belts that snap every 100,000 miles...

Some are submerged in oil and break earlier. They decompose, clog the
oil filter with debris, and the engine is totalled because of oil
failure. Who was the idiot that thought submerging a rubber belt in
oil was a good idea?

Well you are supposed to *change* them, and the oil they run in,
regularly...

Even if you do, they break way before their due time.

It is a known issue, well published.

So far that seems to be mostly a European thing.

Puretech.

You can read the articles with google translator:

<https://www.xataka.com/movilidad/correas-banadas-aceite-hundieron-completo-reputacion-stellantis-a-veces-olvidamos-que-no-fue-unica>

<https://mlin.es/2024/10/03/marcas-de-coches-con-correa-de-distribucion-banada-en-aceite/>

<https://www.autopista.es/noticias-motor/todos-coches-afectados-ya-por-problemas-motor-causa-correas-distribucion-en-aceite_306280_102.html>

English:

<https://www.reddit.com/r/peugeot/comments/18wtur8/is_the_puretech_12_engine_as_notoriously_bad_as/>

<https://today.rtl.lu/news/business-and-tech/a/2208966.html>

<https://www.tiktok.com/@life.behind.cars/video/7523105415056362774>
--
Cheers, Carlos.
ESEfc-Efc+, EUEfc-Efc|;
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-15 01:31, Charlie Gibbs wrote:

We used to have vehicle inspection stations here in B.C. but the
provincial government shut them down around 1982 as a "cost-saving
measure".

The inspection system here (Spain) charges the owners, and they are
mandatory. They make a good profit.

The first inspection is at 4 years of age, then every two years, then
every year.

They check a lot of things. Gases, of course. Transmission, brakes, suspension, lights, safety belts...
--
Cheers, Carlos.
ESEfc-Efc+, EUEfc-Efc|;
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.os.linux.misc

On 2025-10-15, rbowman <bowman@montana.com> wrote:

On Tue, 14 Oct 2025 23:31:45 GMT, Charlie Gibbs wrote:

I wish the inspection stations were still around. Most modern cars
would fail due to overly-bright headlights aimed too high. (Gotta dazzle
the oncoming traffic, doncha know - the car ads say it's cool.) When we
cross the border into Washington I notice that headlights down there
aren't nearly as obnoxious. In fact, the last time I bought a
replacement headlamp I noticed that a number of them were marked "Legal
in Canada and Mexico only". Apparently this is something the States got
right.

The HID systems are really annoying to say nothing of pickups with headlights at eye level if you're in a passenger car. I had cataracts removed from both eyes last year. Prior to that I was mostly flying blind with oncoming headlights.

I've had cataract surgery on one eye; the other one is OK so far,
although I suspect it's just about time to have it done. (Maybe after
my upcoming knee surgery - I'm really getting into "spare parts" mode
these days.) The difference in colour between the eyes is kind of
weird but not too noticeable.

These days when driving at night I often pull my sun visor down
until all I can see of the car ahead is its bumper and wheels.
That nicely blocks light from oncoming cars. If someone with
bright headlights (HID, LED, misadjusted halogens, it's all the
same) is behind me, I adjust my side mirror to direct the beam
back into his own eyes. Not only is it poetic justice, it
discourages tailgating.

I don't take the government too seriously when it comes to road
safety. There are too many inconsistencies, and probably not a
few kickbacks.

I'm not big on gadgetry and my Toyota doesn't have remote entry, lane keeping, and all that but one thing it does have is automatic headlight dimming. In almost all cases it dims the lights when I would manually
either overtaking or approaching. One quirk is a 20 mph curve on my way
home that has a lot of reflective signage that will trigger dimming.

iirc Cadillac had something like that about 50 years ago that didn't work very well. I assume this system has more intelligence and better sensors.

Interesting. But it's one more gadget I don't need. I don't find it
that big a hardship to dim my own headlights. The one gadget I've wanted
for years - and it wouldn't be difficult to implement - is turn signals
that automatically cancel after 30 seconds if the car is in motion.
Or at least click loudly enough to be heard, possibly with a loud
alarm after that 30 seconds.
--
/~\ Charlie Gibbs | Growth for the sake of
\ / <cgibbs@kltpzyxm.invalid> | growth is the ideology
X I'm really at ac.dekanfrus | of the cancer cell.
/ \ if you read it the right way. | -- Edward Abbey
--- Synchronet 3.21a-Linux NewsLink 1.2