On Fri, 20 May 2022 14:03:11 -0400, Arne Vajhøj wrote:

> But note that their existence is optional, if the system does not have
> 32 bit integers then they don't exist.

I have a standard 'typedefs' file that I use for all my C programs. For
example:

typedef int INT,*PINT;

At least I can make adjustments easily, although one has to be careful in
lots of other places (e.g. shifts).

--
My posts are my copyright and if @diy_forums or Home Owners' Hub
wish to copy them they can pay me £1 a message.
Use the BIG mirror service in the UK: http://www.mirrorservice.org
*lightning surge protection* - a w_tom conductor

On 2022-05-20 17:44:50 +0000, Simon Clubley said:

> On 2022-05-20, Arne Vajhøj <arne@vajhoej.dk> wrote:
>>
>> Neither Pascal nor C have well defined types in the standard.
>>
>> The size of integer in bits and one or two complement is implementation
>> specific.
>>

That's true in C until C2X, with the current draft requiring two's
complement, and including type-generic functions for integer overflow
checks.

>
> These days, I avoid this problem in my C code by using the
> uint[8/16/32]_t (and friends) data types.
>

Same here, where I need to care that much about the integer types.

--
Pure Personal Opinion | HoffmanLabs LLC

On 5/20/2022 1:38 PM, Simon Clubley wrote:
> On 2022-05-20, Dave Froble <davef@tsoft-inc.com> wrote:
>> On 5/20/2022 8:16 AM, Arne Vajhøj wrote:
>>
>>> If we go over to VMS Basic, then VSI can easily add new
>>> types and new functions. No problem. But could VSI
>>> remove the variable name suffix does implicit typing
>>> mechanism? I don't think so.
>>>
>>> Arne
>>>
>>
>> Why would you want to remove something that perhaps some users want to continue
>> to use? Leaving it doesn't hurt, and users can choose to not use it.
>>
>
> Because you want to remove dangerous features to force people to write
> more robust code.

There are words to describe that type of thinking. "Tyrant" comes to mind.

Should we do away with parachutes so people cannot sky dive?

Should we do away with cars so people cannot crash?

--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: davef@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486

On 5/20/2022 2:23 PM, Johnny Billquist wrote:
> On 2022-05-20 20:03, Arne Vajhøj wrote:
>> On 5/20/2022 1:44 PM, Simon Clubley wrote:
>>> On 2022-05-20, Arne Vajhøj <arne@vajhoej.dk> wrote:
>>>> Neither Pascal nor C have well defined types in the standard.
>>>>
>>>> The size of integer in bits and one or two complement is
>>>> implementation specific.
>>>
>>> These days, I avoid this problem in my C code by using the
>>> uint[8/16/32]_t
>>> (and friends) data types.
>>
>> Which is as good as you can do in C.
>>
>> But note that their existence is optional, if the system does
>> not have 32 bit integers then they don't exist.
>
> Well, if the system don't have 32 bit integers, then the code is doomed
> to fail anyway, and it's good that it fails at compilation.

If 32 bit is truly required then yes.

But if the system has 36 bit int's and they would work then it is
a different story.

>
>> And intN_t is guaranteed to be two's complement but uintN_T is not.
>
> I would hope that an uintN_t is guaranteed to *not* be two complement.
> It would sortof defeat the meaning of the 'u' if it was two complement.

Egg on my face.

Of course unsigned does not specify how negative numbers are
represented.

Arne

On Friday, May 20, 2022 at 4:26:54 PM UTC-7, Arne Vajhøj wrote:
> > Well, if the system don't have 32 bit integers, then the code is doomed
> > to fail anyway, and it's good that it fails at compilation.
> If 32 bit is truly required then yes.
>
> But if the system has 36 bit int's and they would work then it is
> a different story.
> >
> >> And intN_t is guaranteed to be two's complement but uintN_T is not.
> >
> > I would hope that an uintN_t is guaranteed to *not* be two complement.
> > It would sortof defeat the meaning of the 'u' if it was two complement.
> Egg on my face.
>
> Of course unsigned does not specify how negative numbers are
> represented.

You might enjoy this 2018 CppCon talk by JF Bastien at Apple about how C++ actually works in practice and the effort required to finally guarantee signed integers to be two's complement there so that the compiler will generate the code that you'd hoped that it would generate but it actually doesn't sometimes: https://youtu.be/JhUxIVf1qok

C++ is ugly enough that the only accepted way to perform some of these numeric casts safely (and I've seen this in Firefox code and elsewhere) is to memcpy() from the source to the destination variable. The compiler hopefully has memcpy() as a builtin, recognizes what's going on, and generates the correct cast without even moving the variable between registers (hopefully). Otherwise, it could do the wrong thing. It's been a while since I watched the talk, and so the details have thankfully vanished from my memory.

On 2022-05-20, Stephen Hoffman <seaohveh@hoffmanlabs.invalid> wrote:
> On 2022-05-20 17:44:50 +0000, Simon Clubley said:
>>
>> These days, I avoid this problem in my C code by using the
>> uint[8/16/32]_t (and friends) data types.
>>
>
> Same here, where I need to care that much about the integer types.
>

Person who likes Ada here, remember ? :-)

I always care about my types... :-)

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.

On 2022-05-20, Dave Froble <davef@tsoft-inc.com> wrote:
> On 5/20/2022 1:38 PM, Simon Clubley wrote:
>> On 2022-05-20, Dave Froble <davef@tsoft-inc.com> wrote:
>>>
>>> Why would you want to remove something that perhaps some users want to continue
>>> to use? Leaving it doesn't hurt, and users can choose to not use it.
>>>
>>
>> Because you want to remove dangerous features to force people to write
>> more robust code.
>
> There are words to describe that type of thinking. "Tyrant" comes to mind.
>
> Should we do away with parachutes so people cannot sky dive?
>

Parachutes are a required feature so people don't die.

You also make sure they are built to required safety standards.

> Should we do away with cars so people cannot crash?
>

No, but you ban the ones that are no longer safe.

Likewise, there are very good reasons why Hollerith constants are no
longer in the current versions of Fortran.

Some reading:

https://en.wikipedia.org/wiki/Hollerith_constant

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.

On 5/23/2022 8:34 AM, Simon Clubley wrote:
> On 2022-05-20, Dave Froble <davef@tsoft-inc.com> wrote:
>> On 5/20/2022 1:38 PM, Simon Clubley wrote:
>>> On 2022-05-20, Dave Froble <davef@tsoft-inc.com> wrote:
>>>>
>>>> Why would you want to remove something that perhaps some users want to continue
>>>> to use? Leaving it doesn't hurt, and users can choose to not use it.
>>>>
>>>
>>> Because you want to remove dangerous features to force people to write
>>> more robust code.
>>
>> There are words to describe that type of thinking. "Tyrant" comes to mind.
>>
>> Should we do away with parachutes so people cannot sky dive?
>>
>
> Parachutes are a required feature so people don't die.

If people would not jump out of perfectly airworthy aircraft, they would not be
required.

> You also make sure they are built to required safety standards.

That would be the reserve parachute, for when the primary doesn't open.

>> Should we do away with cars so people cannot crash?
>>
>
> No, but you ban the ones that are no longer safe.

When a fully loaded tri-axle truck runs over a car, then no car is safe.

Thing is Simon, there is no true safety. There is due diligence, and being
careful. Still, shot happens.

What safety is there when that comet plows into (where did you say you lived?)
the planet?

What safety is there in programming when you remove the need for a programmer to
exercise due diligence and careful practices?

Yes, I use tools to evaluate a compiler listing, though I'd rather the compiler
just tell me.

There is a difference between "enforce" and "inform". I'm not into enforcement.

--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: davef@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486

On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>
> What safety is there in programming when you remove the need for a programmer to
> exercise due diligence and careful practices?
>

With that attitude David, you should switch to using C... :-)

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.

On 5/23/2022 1:46 PM, Simon Clubley wrote:
> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>
>> What safety is there in programming when you remove the need for a programmer to
>> exercise due diligence and careful practices?
>>
>
> With that attitude David, you should switch to using C... :-)
>
> Simon.
>

What attitude is that?

Think I'll stick to Basic. Does what I need. Has strings built in.

--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: davef@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486

On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:

> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>
>> What safety is there in programming when you remove the need for a
>> programmer to exercise due diligence and careful practices?
>>
>>
> With that attitude David, you should switch to using C... :-)

C is for wimps. Try BCPL.

--
My posts are my copyright and if @diy_forums or Home Owners' Hub
wish to copy them they can pay me £1 a message.
Use the BIG mirror service in the UK: http://www.mirrorservice.org
*lightning surge protection* - a w_tom conductor

On 5/23/22 16:24, Bob Eager wrote:
> On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
>
>> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>>
>>> What safety is there in programming when you remove the need for a
>>> programmer to exercise due diligence and careful practices?
>>>
>>>
>> With that attitude David, you should switch to using C... :-)
>
> C is for wimps. Try BCPL.
>
>

BCPL is for wimps. Stick with Macro-11.

bill

On Mon, 23 May 2022 18:23:11 -0400, Bill Gunshannon wrote:

> On 5/23/22 16:24, Bob Eager wrote:
>> On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
>>
>>> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>>>
>>>> What safety is there in programming when you remove the need for a
>>>> programmer to exercise due diligence and careful practices?
>>>>
>>>>
>>> With that attitude David, you should switch to using C... :-)
>>
>> C is for wimps. Try BCPL.
>>
>>
>>
>
> BCPL is for wimps. Stick with Macro-11.
>
> bill

I did a BCPL compiler to generate Macro-11. Well, binary instructions.

--
My posts are my copyright and if @diy_forums or Home Owners' Hub
wish to copy them they can pay me £1 a message.
Use the BIG mirror service in the UK: http://www.mirrorservice.org
*lightning surge protection* - a w_tom conductor

On Monday, May 23, 2022 at 9:24:36 PM UTC+1, Bob Eager wrote:
> On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
>
> > On 2022-05-23, Dave Froble <da...@tsoft-inc.com> wrote:
> >>
> >> What safety is there in programming when you remove the need for a
> >> programmer to exercise due diligence and careful practices?
> >>
> >>
> > With that attitude David, you should switch to using C... :-)
> C is for wimps. Try BCPL.
> --
> My posts are my copyright and if @diy_forums or Home Owners' Hub
> wish to copy them they can pay me £1 a message.
> Use the BIG mirror service in the UK: http://www.mirrorservice.org
> *lightning surge protection* - a w_tom conductor

I preferred BCPL to BASIC on the BBC Computer.

On Tue, 2022-05-24 at 01:07 -0700, Ian Miller wrote:
> On Monday, May 23, 2022 at 9:24:36 PM UTC+1, Bob Eager wrote:
> > On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
> >
> > > On 2022-05-23, Dave Froble <da...@tsoft-inc.com> wrote:
> > > >
> > > > What safety is there in programming when you remove the need
> > > > for a
> > > > programmer to exercise due diligence and careful practices?
> > > >
> > > >
> > > With that attitude David, you should switch to using C... :-)
> > C is for wimps. Try BCPL.
> > --
> > My posts are my copyright and if @diy_forums or Home Owners' Hub
> > wish to copy them they can pay me £1 a message.
> > Use the BIG mirror service in the UK: http://www.mirrorservice.org 
> > *lightning surge protection* - a w_tom conductor
>
> I preferred BCPL to BASIC on the BBC Computer.

I preferred 6502 assembler to BASIC on the Beeb.
--
Tactical Nuclear Kittens

In article <jf2fqfFiphjU1@mid.individual.net>,
Bill Gunshannon <bill.gunshannon@gmail.com> wrote:
>On 5/23/22 16:24, Bob Eager wrote:
>> On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
>>
>>> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>>>
>>>> What safety is there in programming when you remove the need for a
>>>> programmer to exercise due diligence and careful practices?
>>>>
>>>>
>>> With that attitude David, you should switch to using C... :-)
>>
>> C is for wimps. Try BCPL.
>
>BCPL is for wimps. Stick with Macro-11.

Anything other than machine code in octal is for pikers.

- Dan C.

On 5/24/22 09:00, Dan Cross wrote:
> In article <jf2fqfFiphjU1@mid.individual.net>,
> Bill Gunshannon <bill.gunshannon@gmail.com> wrote:
>> On 5/23/22 16:24, Bob Eager wrote:
>>> On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
>>>
>>>> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>>>>
>>>>> What safety is there in programming when you remove the need for a
>>>>> programmer to exercise due diligence and careful practices?
>>>>>
>>>>>
>>>> With that attitude David, you should switch to using C... :-)
>>>
>>> C is for wimps. Try BCPL.
>>
>> BCPL is for wimps. Stick with Macro-11.
>
> Anything other than machine code in octal is for pikers.
>

Learned to do that on the IBM 1401 when I learned Autocoder.
Efficiency was important in those days and the output from
the Autocoder translator had way too much fluff. Used to
be fun to Compile a utility program written in Autocoder,
look at the output card deck and then reduce it to one or
two cards max.

bill

On 05/24/22 11:17, Single Stage to Orbit wrote:
> On Tue, 2022-05-24 at 01:07 -0700, Ian Miller wrote:
>> On Monday, May 23, 2022 at 9:24:36 PM UTC+1, Bob Eager wrote:
>>> On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
>>>
>>>> On 2022-05-23, Dave Froble<da...@tsoft-inc.com> wrote:
>>>>>
>>>>> What safety is there in programming when you remove the need
>>>>> for a
>>>>> programmer to exercise due diligence and careful practices?
>>>>>
>>>>>
>>>> With that attitude David, you should switch to using C... :-)
>>> C is for wimps. Try BCPL.
>>> --
>>> My posts are my copyright and if @diy_forums or Home Owners' Hub
>>> wish to copy them they can pay me £1 a message.
>>> Use the BIG mirror service in the UK: http://www.mirrorservice.org
>>> *lightning surge protection* - a w_tom conductor
>>
>> I preferred BCPL to BASIC on the BBC Computer.
>
> I preferred 6502 assembler to BASIC on the Beeb.

I preferred 6502 asm and macro 11 to anything else I had access to
at time.

The rest of you all were lucky :-)...

Chris

On 2022-05-21 01:26, Arne Vajhøj wrote:
> On 5/20/2022 2:23 PM, Johnny Billquist wrote:
>> On 2022-05-20 20:03, Arne Vajhøj wrote:
>>> On 5/20/2022 1:44 PM, Simon Clubley wrote:
>>>> On 2022-05-20, Arne Vajhøj <arne@vajhoej.dk> wrote:
>>>>> Neither Pascal nor C have well defined types in the standard.
>>>>>
>>>>> The size of integer in bits and one or two complement is
>>>>> implementation specific.
>>>>
>>>> These days, I avoid this problem in my C code by using the
>>>> uint[8/16/32]_t
>>>> (and friends) data types.
>>>
>>> Which is as good as you can do in C.
>>>
>>> But note that their existence is optional, if the system does
>>> not have 32 bit integers then they don't exist.
>>
>> Well, if the system don't have 32 bit integers, then the code is
>> doomed to fail anyway, and it's good that it fails at compilation.
>
> If 32 bit is truly required then yes.
>
> But if the system has 36 bit int's and they would work then it is
> a different story.

Well. There is no way to tell if a 36 bit integer would actually be
acceptable if you try to use something you think are 32 bits, so it's
still the safe option to fail such code.

>>> And intN_t is guaranteed to be two's complement but uintN_T is not.
>>
>> I would hope that an uintN_t is guaranteed to *not* be two complement.
>> It would sortof defeat the meaning of the 'u' if it was two complement.
>
> Egg on my face.

:-D

Johnny

On 2022-05-23 14:29, Simon Clubley wrote:
> On 2022-05-20, Stephen Hoffman <seaohveh@hoffmanlabs.invalid> wrote:
>> On 2022-05-20 17:44:50 +0000, Simon Clubley said:
>>>
>>> These days, I avoid this problem in my C code by using the
>>> uint[8/16/32]_t (and friends) data types.
>>>
>>
>> Same here, where I need to care that much about the integer types.
>>
>
> Person who likes Ada here, remember ? :-)
>
> I always care about my types... :-)

In a sense, this is something Ada really got right. You don't declare
how many bits, signedness and so on you want. You declare your integers
with the range you need them to have, and let the language/compiler
figure out how to represent and store them.
And two "integer" types with different ranges are not the same type, or
even closely related, and you cannot just assign from one type to another.

However, if you want to interface with hardware, it becomes a little
messy, since then you do want to specify things in number of bits and
bytes. Since that's what the hardware have.

Johnny

On 2022-05-24 15:00, Dan Cross wrote:
> In article <jf2fqfFiphjU1@mid.individual.net>,
> Bill Gunshannon <bill.gunshannon@gmail.com> wrote:
>> On 5/23/22 16:24, Bob Eager wrote:
>>> On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
>>>
>>>> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>>>>
>>>>> What safety is there in programming when you remove the need for a
>>>>> programmer to exercise due diligence and careful practices?
>>>>>
>>>>>
>>>> With that attitude David, you should switch to using C... :-)
>>>
>>> C is for wimps. Try BCPL.
>>
>> BCPL is for wimps. Stick with Macro-11.
>
> Anything other than machine code in octal is for pikers.

Octal? You had octal? You lucky bastards. We had to be content with binary.

(Go ahead. Make my day. Continue this thread... :-D )

Johnny

On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
> On 5/23/2022 1:46 PM, Simon Clubley wrote:
>> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>>
>>> What safety is there in programming when you remove the need for a programmer to
>>> exercise due diligence and careful practices?
>>>
>>
>> With that attitude David, you should switch to using C... :-)
>>
>
> What attitude is that?
>

Some C programmers say that there's nothing wrong with C and that it's
your fault if you are not careful enough when writing C code.

I am not one of them because, for me, I continue to use C because it's
the most viable language for some tasks, not because it is the best
language for those tasks.

> Think I'll stick to Basic. Does what I need. Has strings built in.
>

It protects you better than C at the expense of not been able to do
everything C does but there are also other languages that help protect
you while still allowing you to do the things that C lets you do.

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.

On 2022-05-24, Johnny Billquist <bqt@softjar.se> wrote:
> On 2022-05-24 15:00, Dan Cross wrote:
>> In article <jf2fqfFiphjU1@mid.individual.net>,
>> Bill Gunshannon <bill.gunshannon@gmail.com> wrote:
>>> On 5/23/22 16:24, Bob Eager wrote:
>>>> On Mon, 23 May 2022 17:46:16 +0000, Simon Clubley wrote:
>>>>
>>>>> On 2022-05-23, Dave Froble <davef@tsoft-inc.com> wrote:
>>>>>>
>>>>>> What safety is there in programming when you remove the need for a
>>>>>> programmer to exercise due diligence and careful practices?
>>>>>>
>>>>>>
>>>>> With that attitude David, you should switch to using C... :-)
>>>>
>>>> C is for wimps. Try BCPL.
>>>
>>> BCPL is for wimps. Stick with Macro-11.
>>
>> Anything other than machine code in octal is for pikers.
>
> Octal? You had octal? You lucky bastards. We had to be content with binary.
>
> (Go ahead. Make my day. Continue this thread... :-D )
>

Lucky you! I had to be happy with using a really small magnet to
program the core memory...

Beat that!

Simon.

PS: On a more serious note, the only time I have seen core memory is
in a museum or in pictures. The thing that strikes me about it is
that it was a destructive read. I can't think of anything else where
reading it destroys the contents. Does anyone know of such technology ?

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.

On 2022-05-24, Single Stage to Orbit <alex.buell@munted.eu> wrote:
> On Tue, 2022-05-24 at 01:07 -0700, Ian Miller wrote:
>>
>> I preferred BCPL to BASIC on the BBC Computer.
>
> I preferred 6502 assembler to BASIC on the Beeb.

It's been a very long time since I used either, but I was left with
the memory that I preferred Z80 over 6502, although the details have
been lost to time...

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.

On 05/24/22 18:41, Simon Clubley wrote:
> On 2022-05-24, Single Stage to Orbit<alex.buell@munted.eu> wrote:
>> On Tue, 2022-05-24 at 01:07 -0700, Ian Miller wrote:
>>>
>>> I preferred BCPL to BASIC on the BBC Computer.
>>
>> I preferred 6502 assembler to BASIC on the Beeb.
>
> It's been a very long time since I used either, but I was left with
> the memory that I preferred Z80 over 6502, although the details have
> been lost to time...
>
> Simon.
>

What made the 6502 a bit special was the variety of addressing modes,
which included pre and post indexed addressing. Very useful for
traversing lists and trees. That and instruction mnemonics
borrowed from pdp11, even if the architecture was a single 8 bit
accumulator + 2 index registers. Not even close to the PDP11 of
the time, but probably the best they could do with the process
technology of the day..

Chris