Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.os.linux.misc > #91099 > unrolled thread

OT: Does Computer Science Need Computers?

Back to article view | Back to comp.os.linux.misc

Contents

  OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-04 02:20 +0000
    Re: OT: Does Computer Science Need Computers? Leroy H <lh@somewhere.net> - 2026-09-04 03:17 +0000
      Re: OT: Does Computer Science Need Computers? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2026-09-04 12:56 +0000
        Re: OT: Does Computer Science Need Computers? The Natural Philosopher <tnp@invalid.invalid> - 2026-09-04 14:44 +0100
          Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-04 18:48 +0000
        Re: OT: Does Computer Science Need Computers? Leroy H <lh@somewhere.net> - 2026-09-04 16:00 +0000
          Re: OT: Does Computer Science Need Computers? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2026-09-04 23:19 +0000
            Re: OT: Does Computer Science Need Computers? Leroy H <lh@somewhere.net> - 2026-09-05 00:59 +0000
            Re: OT: Does Computer Science Need Computers? Leroy H <lh@somewhere.net> - 2026-09-05 01:26 +0000
              Intel and IEEE 754 (was: Re: OT: Does Computer Science Need Computers?) Nuno Silva <nunojsilva@invalid.invalid> - 2026-09-05 03:06 +0100
                Re: Intel and IEEE 754 (was: Re: OT: Does Computer Science Need Computers?) Leroy H <lh@somewhere.net> - 2026-09-05 02:27 +0000
              Re: OT: Does Computer Science Need Computers? c186282 <c186282@nnada.net> - 2026-09-05 03:25 -0400
                Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-05 17:56 +0000
            Re: OT: Does Computer Science Need Computers? Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-09-05 21:30 +0000
              Re: OT: Does Computer Science Need Computers? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2026-09-05 22:38 +0000
                Re: OT: Does Computer Science Need Computers? Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-09-06 01:59 +0000
                Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-06 03:08 +0000
                  Re: OT: Does Computer Science Need Computers? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2026-09-06 03:51 +0000
                    Re: OT: Does Computer Science Need Computers? Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-09-06 06:43 +0000
              Re: OT: Does Computer Science Need Computers? Leroy H <lh@somewhere.net> - 2026-09-06 02:43 +0000
        Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-04 18:37 +0000
          Re: OT: Does Computer Science Need Computers? c186282 <c186282@nnada.net> - 2026-09-04 21:20 -0400
            Re: OT: Does Computer Science Need Computers? Leroy H <lh@somewhere.net> - 2026-09-05 01:40 +0000
            Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-05 03:46 +0000
              Re: OT: Does Computer Science Need Computers? c186282 <c186282@nnada.net> - 2026-09-05 03:36 -0400
                Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-05 18:09 +0000
                  Re: OT: Does Computer Science Need Computers? The Natural Philosopher <tnp@invalid.invalid> - 2026-09-05 19:23 +0100
                    Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-06 02:49 +0000
                      Re: OT: Does Computer Science Need Computers? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2026-09-06 03:51 +0000
        Re: OT: Does Computer Science Need Computers? Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-09-04 22:36 +0000
        Re: OT: Does Computer Science Need Computers? Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-09-06 06:50 +0000
          Re: OT: Does Computer Science Need Computers? c186282 <c186282@nnada.net> - 2026-09-07 04:14 -0400
            Re: OT: Does Computer Science Need Computers? The Natural Philosopher <tnp@invalid.invalid> - 2026-09-07 12:05 +0100
              Re: OT: Does Computer Science Need Computers? Chris Ahlstrom <OFeem1987@teleworm.us> - 2026-09-07 09:17 -0400
                Re: OT: Does Computer Science Need Computers? The Natural Philosopher <tnp@invalid.invalid> - 2026-09-07 15:02 +0100
                Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-07 19:16 +0000
                  Re: OT: Does Computer Science Need Computers? The Natural Philosopher <tnp@invalid.invalid> - 2026-09-07 22:26 +0100
                    Re: OT: Does Computer Science Need Computers? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2026-09-08 04:18 +0000
                      Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-08 04:38 +0000
                      Re: OT: Does Computer Science Need Computers? Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-09-08 05:43 +0000
                        Re: OT: Does Computer Science Need Computers? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2026-09-09 04:25 +0000
                          Re: OT: Does Computer Science Need Computers? The Natural Philosopher <tnp@invalid.invalid> - 2026-09-09 09:35 +0100
                  Re: OT: Does Computer Science Need Computers? Chris Ahlstrom <OFeem1987@teleworm.us> - 2026-09-07 18:58 -0400
                    Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-08 04:47 +0000
              Re: OT: Does Computer Science Need Computers? Rich <rich@example.invalid> - 2026-09-14 17:18 +0000
    Re: OT: Does Computer Science Need Computers? Nuno Silva <nunojsilva@invalid.invalid> - 2026-09-05 02:50 +0100
      Re: OT: Does Computer Science Need Computers? The Natural Philosopher <tnp@invalid.invalid> - 2026-09-05 08:37 +0100
        Re: OT: Does Computer Science Need Computers? Nuno Silva <nunojsilva@invalid.invalid> - 2026-09-05 10:20 +0100
          Re: OT: Does Computer Science Need Computers? rbowman <bowman@montana.com> - 2026-09-05 18:31 +0000

Page 1 of 3  [1] 2 3  Next page →

#91099 — OT: Does Computer Science Need Computers?

https://www.quantamagazine.org/does-computer-science-need-computers-20260828/

[toc] | [next] | [standalone]

#91100

On 4 Sep 2026 02:20:42 GMT, rbowman wrote:

> https://www.quantamagazine.org/does-computer-science-need-computers-20260828/
>

Bullshit fluff.

It was the need for calculating speed that drove the development of
the "computer."

The first "computers" were veritable armies of women who were enlisted
to perform calculations but, needless to say, this was terribly slow.

Then there came the analog computer but these were limited to specific
functions like the Norden bomb sight.

Digital computers allowed general purpose computing and their development
was spurred, after Bletchley Park, by calculating neutron diffusion in
the nascent fission/fusion bombs.

The need for speed, and nothing else, compelled the development of the digital
computer.

But the theory of computation has grown around the concept of the digital
computer.  As the article indicates, the early ideas of computation (e.g.
George Boole, Brouwer, Norbert Wiener),  were motivated by the logical
foundations of mathematics.  But current practice is essentially detached
from this basis.

Does computer science need computers?  This is just abominable click bait.

What we now practice as "computer science" is totally grounded in computers,
be they digital, ternary, or something else.

[toc] | [prev] | [next] | [standalone]

#91102

On 2026-09-04, Leroy H <lh@somewhere.net> wrote:

> Does computer science need computers?  This is just abominable click bait.
>
> What we now practice as "computer science" is totally grounded in computers,
> be they digital, ternary, or something else.

I dropped out of computer science because it was too theoretical.
At the time, computer science was a subsidiary of the math
department; it didn't become a department in its own right
until after I was gone.

Most CS dweebs spent their time sitting in an ivory tower
contemplating the whichness of what; mention COBOL (or any
real-world application, for that matter) and their faces
would turn green.

I left to find a programming job in the Real World [tm] instead.

-- 
/~\  Charlie Gibbs                  |  In this world there are
\ /  <cgibbs@kltpzyxm.invalid>      |  two kinds of people:
 X   I'm really at ac.dekanfrus     |  1. Those who can extrapolate
/ \  if you read it the right way.  |  from incomplete data.

[toc] | [prev] | [next] | [standalone]

#91103

On 04/09/2026 13:56, Charlie Gibbs wrote:
> On 2026-09-04, Leroy H <lh@somewhere.net> wrote:
> 
>> Does computer science need computers?  This is just abominable click bait.
>>
>> What we now practice as "computer science" is totally grounded in computers,
>> be they digital, ternary, or something else.
> 
> I dropped out of computer science because it was too theoretical.
> At the time, computer science was a subsidiary of the math
> department; it didn't become a department in its own right
> until after I was gone.
> 
> Most CS dweebs spent their time sitting in an ivory tower
> contemplating the whichness of what; mention COBOL (or any
> real-world application, for that matter) and their faces
> would turn green.
> 
> I left to find a programming job in the Real World [tm] instead.
> 
Amen to that.

Eventually I discovered that what I had was not knowledge of computer 
science, but software engineering which is how you actually produce 
software on time, on budget, and tested.

CompSci has its place, but it's not an especially important one. Its 
mainly valuable in fundamental algorithm development or encryption schemata.


-- 
"What do you think about Gay Marriage?"
"I don't."
"Don't what?"
"Think about Gay Marriage."

[toc] | [prev] | [next] | [standalone]

#91106

On Fri, 4 Sep 2026 14:44:35 +0100, The Natural Philosopher wrote:

> Eventually I discovered that what I had was not knowledge of computer
> science, but software engineering which is how you actually produce
> software on time, on budget, and tested.

Good technical writers, while not expert in the particular field, can take 
geek speak and make it comprehensible. Programming is similar. The old 
'jack of all trades, master of none' comes to mind except the mastery is 
the ability to grasp what is needed in a particular trade and make it 
happen.

[toc] | [prev] | [next] | [standalone]

#91104

On Fri, 04 Sep 2026 12:56:59 GMT, Charlie Gibbs wrote:

> 
> I left to find a programming job in the Real World [tm] instead.
>

You are then only using the algorithmic tools that were developed by
generations of "Ivory Tower" computer scientists.

Would you have been able to apply space filling curves to database
searches:

<https://towardsdatascience.com/spatial-index-space-filling-curves-e67baec2186a/>

Would you have been able to develop the methods of dynamic programming:

<https://stackoverflow.blog/2022/01/31/the-complete-beginners-guide-to-dynamic-programming/>

Would you have been able to develop the critical techniques of representing
and processing floating point numbers:

<https://arxiv.org/html/2012.02492v1>

A knowledge of programming (i.e. coding) only will leave one with
only brute force and highly inefficient means of dealing with complex
problems.

But it's probably all moot.  Most programming nowadays is done
using frameworks which essentially do the programming for the
"programmer."  Vibe coding  is only a small step up from this level.

[toc] | [prev] | [next] | [standalone]

#91108

On 2026-09-04, Leroy H <lh@somewhere.net> wrote:

> On Fri, 04 Sep 2026 12:56:59 GMT, Charlie Gibbs wrote:
>
>> I left to find a programming job in the Real World [tm] instead.
>
> You are then only using the algorithmic tools that were developed by
> generations of "Ivory Tower" computer scientists.

Not a lot of them, even.  I was happiest in assembly language.

> Would you have been able to apply space filling curves to database
> searches:
>
> <https://towardsdatascience.com/spatial-index-space-filling-curves-e67baec2186a/>

I've never done that, then or now.

> Would you have been able to develop the methods of dynamic programming:
>
> <https://stackoverflow.blog/2022/01/31/the-complete-beginners-guide-to-dynamic-programming/>

I don't even know what that is.

> Would you have been able to develop the critical techniques of representing
> and processing floating point numbers:
>
> <https://arxiv.org/html/2012.02492v1>

In my 50-plus year career, I can count the number of times I've
used floating point on the fingers of one hand.

> A knowledge of programming (i.e. coding) only will leave one with
> only brute force and highly inefficient means of dealing with complex
> problems.

I've accumulated a lot of knowledge beyond mere coding.
I still shake my head when I see people making design
blunders that I learned to avoid decades ago.

> But it's probably all moot.  Most programming nowadays is done
> using frameworks which essentially do the programming for the
> "programmer."  Vibe coding  is only a small step up from this level.

Somehow I found a career path that avoided many things that
people nowadays see as essential.  I've never been out of
work, though.  That must mean something...

-- 
/~\  Charlie Gibbs                  |  In this world there are
\ /  <cgibbs@kltpzyxm.invalid>      |  two kinds of people:
 X   I'm really at ac.dekanfrus     |  1. Those who can extrapolate
/ \  if you read it the right way.  |  from incomplete data.

[toc] | [prev] | [next] | [standalone]

#91109

On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:

> 
> In my 50-plus year career, I can count the number of times I've
> used floating point on the fingers of one hand.
> 

If you are using an un-typed, high-level language then you
are using floating point without being aware of it.

But I wonder how many programmers know to properly evaluate
a polynomial expression, i.e. the kind of simple equation that
we all doubtless have seen many times in school:

X^n + X^(n-1) + ... + X^3 + X^2 + X + A

There is still comp sci research being done to optimize this.

Eventually, it will end up in the back rooms of Python and
no one will be the wiser.

[toc] | [prev] | [next] | [standalone]

#91111

On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:

> 
> In my 50-plus year career, I can count the number of times I've
> used floating point on the fingers of one hand.
> 

The hardware engineers at Intel should sympathize, for they are
responsible for the biggest floating point blunder of all time --
and it is still present today in every Intel machine.

I don't want to expound the technical details but the nitty-gritty
of it all is found here:

<https://randomascii.wordpress.com/2014/10/09/intel-underestimates-error-bounds-by-1-3-quintillion/>

The result of this major Intel flub is that no math library,
including GNU/Linux GLIBC, uses hardware for trigonometric
calculations.  Rather they use substitute software routines.

The whole point of hardware is to speed things up significantly
but Intel ruined the ball game.

There is/has been no fix.  It would cost Intel trillions of
dollars to replace all the junk hardware.  The best that they
can offer is a footnote in their programming manuals to be wary
of hardware trig functions.

Surprisingly (or perhaps not), this HUGE ISSUE is little known
among programmers.

[toc] | [prev] | [next] | [standalone]

#91114 — Intel and IEEE 754 (was: Re: OT: Does Computer Science Need Computers?)

On 2026-09-05, Leroy H wrote:

> On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:
>
>> 
>> In my 50-plus year career, I can count the number of times I've
>> used floating point on the fingers of one hand.
>> 
>
> The hardware engineers at Intel should sympathize, for they are
> responsible for the biggest floating point blunder of all time --
> and it is still present today in every Intel machine.
>
> I don't want to expound the technical details but the nitty-gritty
> of it all is found here:
>
> <https://randomascii.wordpress.com/2014/10/09/intel-underestimates-error-bounds-by-1-3-quintillion/>
>
> The result of this major Intel flub is that no math library,
> including GNU/Linux GLIBC, uses hardware for trigonometric
> calculations.  Rather they use substitute software routines.
>
> The whole point of hardware is to speed things up significantly
> but Intel ruined the ball game.
>
> There is/has been no fix.  It would cost Intel trillions of
> dollars to replace all the junk hardware.  The best that they
> can offer is a footnote in their programming manuals to be wary
> of hardware trig functions.


From what I'm reading, it's just that they might be off the hook because
fsin (or its precision?) is not specified in the standard. It clearly
isn't the best they can offer, as history can attest.


<gopher://quix.us/0/text/humor/COMPUTER/penthaha.txt>


> Surprisingly (or perhaps not), this HUGE ISSUE is little known
> among programmers.
>

-- 
Nuno Silva

[toc] | [prev] | [next] | [standalone]

#91115 — Re: Intel and IEEE 754 (was: Re: OT: Does Computer Science Need Computers?)

On Sat, 05 Sep 2026 03:06:48 +0100, Nuno Silva wrote:

> 
> From what I'm reading, it's just that they might be off the hook because
> fsin (or its precision?) is not specified in the standard. It clearly
> isn't the best they can offer, as history can attest.
> 

Yes, it is true that IEE754 specifies correctly rounded results only
for arithmetic and square root operations.

But the Intel programming manuals were clearly in error and that is
clearly a fault of Intel's engineers.  As the linked article states,
even Linus Torvalds accepted the original Intel documentation:

<https://gcc.gnu.org/legacy-ml/gcc/2001-08/msg00679.html>

Even software routines don't provide correct trig rounding, at
least not GLIBC.  I have tested GLIBC using the ucbtests:

<https://www.netlib.org/fp/>

The sin() and cos() are off by about a maximum 3 ulps when it should
be 1/2 ulp.

I believe that there is, or soon will be, a new C language standard
that mandates correctly rounded trig and most other transcendental
functions.  GLIBC will incorporate this.

[toc] | [prev] | [next] | [standalone]

#91117

On 9/4/26 21:26, Leroy H wrote:
> On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:
> 
>>
>> In my 50-plus year career, I can count the number of times I've
>> used floating point on the fingers of one hand.
>>
> 
> The hardware engineers at Intel should sympathize, for they are
> responsible for the biggest floating point blunder of all time --
> and it is still present today in every Intel machine.
> 
> I don't want to expound the technical details but the nitty-gritty
> of it all is found here:
> 
> <https://randomascii.wordpress.com/2014/10/09/intel-underestimates-error-bounds-by-1-3-quintillion/>
> 
> The result of this major Intel flub is that no math library,
> including GNU/Linux GLIBC, uses hardware for trigonometric
> calculations.  Rather they use substitute software routines.
> 
> The whole point of hardware is to speed things up significantly
> but Intel ruined the ball game.
> 
> There is/has been no fix.  It would cost Intel trillions of
> dollars to replace all the junk hardware.  The best that they
> can offer is a footnote in their programming manuals to be wary
> of hardware trig functions.
> 
> Surprisingly (or perhaps not), this HUGE ISSUE is little known
> among programmers.

   Intel DID fuck it up - kinda BADLY. There's
   fallout to this day.

   No, you DON'T design a new bridge or building
   or scientific theory using their float shit.

   There are 'software' work-arounds ... but they
   are much slower than using the NPU.

   Pretty fuckin' bad for such a large corp with
   a long history. "Rush To Production" no doubt.

   Now, really, alternate providers like AMD kind
   of "cheat" - see where Intel failed and then do
   their own chips with built-in fixes.

   But, if you need to virtually stress-test a
   new bridge design ........ Intel ??? Just
   what ELSE did they miss ???

[toc] | [prev] | [next] | [standalone]

#91125

On Sat, 5 Sep 2026 03:25:13 -0400, c186282 wrote:

>    Now, really, alternate providers like AMD kind of "cheat" - see where
>    Intel failed and then do their own chips with built-in fixes.

Cheat? Isn't building the better mousetrap what it's all about? 

[toc] | [prev] | [next] | [standalone]

#91130

On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:

> In my 50-plus year career, I can count the number of times I've used
> floating point on the fingers of one hand.

I use floats routinely for times nowadays. Because the most convenient
unit for doing most time calculations is seconds and fractions of a
second, rather than some arbitrary integer unit like milliseconds or
microseconds or whatever.

[toc] | [prev] | [next] | [standalone]

#91132

On 2026-09-05, Lawrence D’Oliveiro <ldo@nz.invalid> wrote:

> On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:
>
>> In my 50-plus year career, I can count the number of times I've used
>> floating point on the fingers of one hand.
>
> I use floats routinely for times nowadays. Because the most convenient
> unit for doing most time calculations is seconds and fractions of a
> second, rather than some arbitrary integer unit like milliseconds or
> microseconds or whatever.

Let me think about that one.  I'll be back in 59.999999 seconds.

-- 
/~\  Charlie Gibbs                  |  In this world there are
\ /  <cgibbs@kltpzyxm.invalid>      |  two kinds of people:
 X   I'm really at ac.dekanfrus     |  1. Those who can extrapolate
/ \  if you read it the right way.  |  from incomplete data.

[toc] | [prev] | [next] | [standalone]

#91136

On Sat, 05 Sep 2026 22:38:17 GMT, Charlie Gibbs wrote:

> On 2026-09-05, Lawrence D’Oliveiro <ldo@nz.invalid> wrote:
>
>> On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:
>>
>>> In my 50-plus year career, I can count the number of times I've
>>> used floating point on the fingers of one hand.
>>
>> I use floats routinely for times nowadays. Because the most
>> convenient unit for doing most time calculations is seconds and
>> fractions of a second, rather than some arbitrary integer unit like
>> milliseconds or microseconds or whatever.
>
> Let me think about that one.  I'll be back in 59.999999 seconds.

Looks like you were a microsecond late. In the meantime, your session
cookie has expired.

(Oh, you didn’t know the authentication server was running a
hard-realtime OS?)

[toc] | [prev] | [next] | [standalone]

#91139

On Sat, 05 Sep 2026 22:38:17 GMT, Charlie Gibbs wrote:

> On 2026-09-05, Lawrence D’Oliveiro <ldo@nz.invalid> wrote:
> 
>> On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:
>>
>>> In my 50-plus year career, I can count the number of times I've used
>>> floating point on the fingers of one hand.
>>
>> I use floats routinely for times nowadays. Because the most convenient
>> unit for doing most time calculations is seconds and fractions of a
>> second, rather than some arbitrary integer unit like milliseconds or
>> microseconds or whatever.
> 
> Let me think about that one.  I'll be back in 59.999999 seconds.

Not 59.9999827? A lot of GIS data points are doubles. That means you have 
to mess around calculating epsilons since 59.999999 obviously isn't equal 
to 59.9999827.  There is an operation called 'snapping'. Essentially 
you're saying 'if the two points are 5 feet apart they're the same damn 
point.' 

https://desktop.arcgis.com/en/arcmap/latest/manage-data/editing-
fundamentals/about-snapping.htm

It's embarrassing when an intended intersection really isn't connected. As 
far as shortest path routing is concerned it isn't routable.

[toc] | [prev] | [next] | [standalone]

#91141

On 2026-09-06, rbowman <bowman@montana.com> wrote:

> On Sat, 05 Sep 2026 22:38:17 GMT, Charlie Gibbs wrote:
>
>> On 2026-09-05, Lawrence D’Oliveiro <ldo@nz.invalid> wrote:
>> 
>>> On Fri, 04 Sep 2026 23:19:28 GMT, Charlie Gibbs wrote:
>>>
>>>> In my 50-plus year career, I can count the number of times I've used
>>>> floating point on the fingers of one hand.
>>>
>>> I use floats routinely for times nowadays. Because the most convenient
>>> unit for doing most time calculations is seconds and fractions of a
>>> second, rather than some arbitrary integer unit like milliseconds or
>>> microseconds or whatever.
>> 
>> Let me think about that one.  I'll be back in 59.999999 seconds.
>
> Not 59.9999827? A lot of GIS data points are doubles. That means you have 
> to mess around calculating epsilons since 59.999999 obviously isn't equal 
> to 59.9999827.  There is an operation called 'snapping'. Essentially 
> you're saying 'if the two points are 5 feet apart they're the same damn 
> point.' 

As the old commercial says, "It takes the worry out of being close."

> https://desktop.arcgis.com/en/arcmap/latest/manage-data/editing-
> fundamentals/about-snapping.htm
>
> It's embarrassing when an intended intersection really isn't connected. As 
> far as shortest path routing is concerned it isn't routable.

And that's why I've done so little floating-point work -
unless you really need it there are all sorts of potential
hassles.  Long integer fields with enough implied decimals
can save a lot of pain.  Packed decimal fields used by the
IBM 360 and clones could be up to 31 digits long - enough
for just about anything.

-- 
/~\  Charlie Gibbs                  |  In this world there are
\ /  <cgibbs@kltpzyxm.invalid>      |  two kinds of people:
 X   I'm really at ac.dekanfrus     |  1. Those who can extrapolate
/ \  if you read it the right way.  |  from incomplete data.

[toc] | [prev] | [next] | [standalone]

#91142

On Sun, 06 Sep 2026 03:51:11 GMT, Charlie Gibbs wrote:

> Long integer fields with enough implied decimals can save a lot of
> pain. Packed decimal fields used by the IBM 360 and clones could be
> up to 31 digits long - enough for just about anything.

Not enough dynamic range.

[toc] | [prev] | [next] | [standalone]

#91137

On Sat, 5 Sep 2026 21:30:10 -0000 (UTC), Lawrence D’Oliveiro wrote:

> 
> I use floats routinely for times nowadays. Because the most convenient
> unit for doing most time calculations is seconds and fractions of a
> second, rather than some arbitrary integer unit like milliseconds or
> microseconds or whatever.
>

I'm not sure what you mean exactly as "time calculations" but the
recommended way of handling these in the C language is with the
extensive time processing routines available with GLIBC: 

<https://sourceware.org/glibc/manual/2.44/html_node/Date-and-Time.html>

Nanosecond resolution is possible.

[toc] | [prev] | [next] | [standalone]

Page 1 of 3  [1] 2 3  Next page →

Back to top | Article view | comp.os.linux.misc

csiph-web