Preamble: I suck at C and I'm proficient in Pascal.

While debugging a piece of C I got into a serious case of pointer mangling.

So, I asked ChatGPT to prepare a table comparing the two.

"please prepare a table comparing pointer data types and pointer
operands between FreePascal and C. Use FreePascal as the starting
point."

Resulted in a nicely prepared table and I quickly debugged my program
(for slow values of quickly).

But that's not it...

Then, on a whim: I asked: "
prepare a comparison of pascal record to C struc where the pascal record
has elements: a, b: word; c:array [1..10] of real; and next which is a
pointer to the record type.

And it did. And it made an error. So I said to it:
"I believe you need to first have a pointer to the MyRecord such as:
MyRecordPtr = ^MyRecord before the record and next should be of type
MyRecordPtr."

ChatGPT replied:
"You are absolutely correct, and I apologize for the oversight. In both
Pascal and C, when you have a pointer to a record or struct within the
same type definition, you should declare a type for that pointer first.
Here's the corrected comparison:"

and put out the correct form (which is - for the case requested):

type
MyRecordPtr = ^MyRecord; // Declare a pointer type
MyRecord = record
a, b: Word;
c: array[1..10] of Real;
next: MyRecordPtr; // Use the pointer type
end;

and for C:

struct MyStruct {
unsigned short a;
unsigned short b;
float c[10];
struct MyStruct* next;
};

(Nasty C letting programmers declare new types and vars on the fly ...
sigh).

--
“Markets can remain irrational longer than your can remain solvent.”
- John Maynard Keynes.

On 15/10/2023 20:51, Alan Browne wrote:
>
> Preamble: I suck at C and I'm proficient in Pascal.
>
> While debugging a piece of C I got into a serious case of pointer mangling.
>
> So, I asked ChatGPT to prepare a table comparing the two.
>
>    "please prepare a table comparing pointer data types and pointer
>     operands between FreePascal and C.  Use FreePascal as the starting
>     point."
>
> Resulted in a nicely prepared table and I quickly debugged my program
> (for slow values of quickly).
>
> But that's not it...
>
> Then, on a whim: I asked: "
> prepare a comparison of pascal record to C struc where the pascal record
> has elements:   a, b: word; c:array [1..10] of real; and next which is a
> pointer to the record type.
>
> And it did.  And it made an error.  So I said to it:
> "I believe you need to first have a pointer to the MyRecord such as:
> MyRecordPtr = ^MyRecord before the record and next should be of type
> MyRecordPtr."
>
> ChatGPT replied:
> "You are absolutely correct, and I apologize for the oversight. In both
> Pascal and C, when you have a pointer to a record or struct within the
> same type definition, you should declare a type for that pointer first.
> Here's the corrected comparison:"
>
> and put out the correct form (which is - for the case requested):
>
> type
>   MyRecordPtr = ^MyRecord;  // Declare a pointer type
>   MyRecord = record
>     a, b: Word;
>     c: array[1..10] of Real;
>     next: MyRecordPtr;  // Use the pointer type
>   end;
>
> and for C:
>
> struct MyStruct {
>   unsigned short a;
>   unsigned short b;
>   float c[10];
>   struct MyStruct* next;
> };
>
> (Nasty C letting programmers declare new types and vars on the fly ...
> sigh).
>
>
Hmmm.
Those declarations do not represent the same types.
[1..10] is not the same as [10].
--
Bill F.

On 2023-10-15 20:56, moi wrote:
> On 15/10/2023 20:51, Alan Browne wrote:

>> and put out the correct form (which is - for the case requested):
>>
>> type
>>    MyRecordPtr = ^MyRecord;  // Declare a pointer type
>>    MyRecord = record
>>      a, b: Word;
>>      c: array[1..10] of Real;
>>      next: MyRecordPtr;  // Use the pointer type
>>    end;
>>
>> and for C:
>>
>> struct MyStruct {
>>    unsigned short a;
>>    unsigned short b;
>>    float c[10];
>>    struct MyStruct* next;
>> };
>>
>> (Nasty C letting programmers declare new types and vars on the fly ...
>> sigh).
>>
>>
> Hmmm.
> Those declarations do not represent the same types.

For Pascal "Real" is platform dependent and can be 32 or 64 bits.

> [1..10] is not the same as [10].

In C it would be 0 based, but in both you get a 10 element array of real
numbers.

--
“Markets can remain irrational longer than your can remain solvent.”
- John Maynard Keynes.