Bad Practice - Mix Int and Long Int

Is it a bad idea to mix integer types ?
Does it affect speed ?
Is it considered a "bad practice" ?
Also, is it terrible to use 64-bit integers only ?

Code:

---

int64_t a = 514915891581; // some large value
int32_t b = 5488;
a = a / b; // implicit conversion - b is converted to int64_t

---

Quote:

---

Is it a bad idea to mix integer types ?

---

It can be. As your example shows, when you mix integer types in arithmetic expressions, the language applies conversion rules, some of which are hard to remember.

Quote:

---

Does it affect speed ?

---

I don't know, but I'm willing to bet you'd be hard-pressed to measure any difference.

Quote:

---

Is it considered a "bad practice" ?

---

The Go programming language forbids it, but only so that they didn't have to define the implicit conversion rules. My approach is to do whatever -Wall -Wextra -pedantic -Werror will allow.

Quote:

---

Also, is it terrible to use 64-bit integers only ?

---

No, it is not terrible. Your 64-bit machine sees the world mostly as 64-bit integers.

The point is, we will always have to deal with a small value,
which results in a mix of 32-bit, 64-bit, and sometimes even 8-bit integers.

Sometimes this helps because it is readable, and in that aspect,
it is easy to know why an 8-bit integer is used.

However this also creates a lack of uniformity, and this can become very ugly,
with 1000+ lines of mixing integers.

Also mixing many integers types, can lead to overflows that can be hard to catch, but does it make sense to have 10 000 lines with 64-bit integers only ?

What is your take ??

Quote:

---

Originally Posted by kodax

The point is, we will always have to deal with a small value,
which results in a mix of 32-bit, 64-bit, and sometimes even 8-bit integers.

---

I don't consider a 64 bit integer, a "small value".

If you know the range of values you will be dealing with, then choose the integer size that will handle all. Look at limits.h.

Are the values all unsigned, or a mix of signed and unsigned?

I would avoid mixing integer sizes without a legitimate reason to do so. If you do, I would avoid downsizing to avoid possible integer overflow without the need for extra code.

I use only signed integers.

The idea was to use 64-bit integers when needed and to stick to the regular 32-bit integers.
It is not a problem and it is not wrong, but it can look quite uncommon to use only 64-bit integers.

Quote:

---

Originally Posted by kodax

I use only signed integers.

The idea was to use 64-bit integers when needed and stick to the regular 32-bit integers.
It is not a problem and it is not wrong, but it can look quite uncommon to use only 64-bit integers.

---

Do any of your signed integers exceed 2,147,483,647? (32 bit INT_MAX) If not, then not need to use 64 bit integers.

What do these values represent?

It does not exceed, it's just that there is a need to use 64-bit integers.

The point was that more than 1000 lines of codes would be mixed with 32-bit and 64-bit integers.
In that case, I wanted to use only 64-bit integers to make things more uniformed.