Make ScLang's integer 64bits

[JordanHendersonMusic]

Purpose and Motivation

More numbers good.

There should be no breaking changes.

This is currently more of a test to see what the CI makes of it.

Types of changes

The slot already had a int64 in, this just updates everything else (mostly function signatures) and tries to keep the old behaviour where possible.

There are many many places where integers are just cast around willy nilly, no attempt is made to fix this here, but should happen in the future — I have tried to make some of these casts explicit.

To-do list

• Code is tested
• All tests are passing
• Updated documentation
• This PR is ready for review

[JordanHendersonMusic]

Does anyone know why the CI failing to pull in the git submodules?

[JordanHendersonMusic]

Just closing this because I no longer have time to play with it.

[telephon]

It is not good to make integer 64 bit because it prevents a possible future optimisation. We should just use integer floats if we need higher numbers.

[smoge]

It is not good to make integer 64 bit because it prevents a possible future optimisation. We should just use integer floats if we need higher numbers.

At least on x86_64, there is no performance difference. It won't get faster nor worse. Although there are 32bit computers around. This change would have to allow different compilations. I think it hasn't been done because it's a massive and repetitive work.

[JordanHendersonMusic]

Just for reference, here are some odd things I found...

1. Int already is 64bit on 64bit machines
   

   supercollider/lang/LangSource/PyrSlot64.h

   Line 57 in 28a0b12

   int64 i;

2. So is char?
   

   supercollider/lang/LangSource/PyrSlot64.h

   Line 56 in 28a0b12

   int64 c; /* char */

3. 32bit is enforced by casting through the getters and setter for the slot, but only in some cases on some architectures
   32bit setter
   

   supercollider/lang/LangSource/PyrSlot64.h

   Line 165 in 28a0b12

   inline void SetRaw(PyrSlot* slot, int val) {

   
   This setter seems problematic because long is 64 on unix and 32 on windows (https://en.cppreference.com/w/cpp/language/types)
   

   supercollider/lang/LangSource/PyrSlot64.h

   Line 169 in 28a0b12

   inline void SetRaw(PyrSlot* slot, long val) {

4. Bit shifting is weird

1 << 30 == 1073741824 // correct
1 << 31 == -2147483648 // correct but (I think) undefined behaviour according to the C++ spec
1 << 31 - 1 == 2147483647 // correct
1 << 36 == 0 // okay just sets higher bits to zero, standard C rules
1 << 60 == 0 
1 << 64  == 1 /// hmmm....
1 << 65  == 2

I think this is because of the long issue or some odd implicit casting, and might not occur on windows?
Left shift uses long

supercollider/lang/LangSource/PyrMathOps.cpp

Line 678 in 28a0b12

long ia = slotRawInt(a);

[smoge]

1 << 30 == 1073741824 // correct
1 << 31 == -2147483648 // correct but (I think) undefined behaviour according to the C++ spec
1 << 31 - 1 == 2147483647 // correct
1 << 36 == 0 // okay just sets higher bits to zero, standard C rules
1 << 60 == 0

This is the expected behavior. Tthe leftmost bit in a signed integer represents its sign

EDIT: I wonder what further bitwise shifting is actually doing. Is it specific to sclang or also cpp?

(Maybe there was some work to transition to int64? Would be informative to see the git dates)