Develop Reference

Is it good practice to use appropriate data types?

This might be a stupid question but here goes:
I'm unsure whether it's habit, laziness, or the correct thing to do. But I find myself using data types that I simply don't need to use, for the sake of consistency, and making things a little bit easier to code later on (i.e casting data types).
Let me show you an example:
public int Number1 { get; set; }
Number1 is just a regular whole number, but imagine the scenario where it will always (or has to be) be a positive number, so I could in fact use uint instead of an int.
However, one of the problems with using lots of different data types is the risk of having to use explicit casting, so I would have lots of noisy code. Using common data types such as int helps to reduce the noise but would this have any effect on the overall performance of an application compared to casting?
int one = 2;
uint two = 1;
one = (int)two;
This noise isn't necessarily bad, but it can be frustrating. There could also be an issue where the uint could be higher than the maximum value of an int.
I find that in many examples, int seems to be a very commonly used data type for working with whole numbers, I very rarely see other data types being used. Why is this?
Is there anything wrong with using 'incorrect' data types, is it more efficient to use an int, rather than a uint, or a short ect?
Thanks in advance.

The best practice is to not simply use signed or unsigned numbers based on the nature of the entities that you want to represent, but to actually declare an entirely new data type for each different type of entity that you are dealing with.
So, for example, a temperature requires a signed number, but the best practice is to declare a new "Temperature" value type. (struct Temperature)
The beauty of C# is that:
If your struct Temperature contains nothing but a float, it will occupy exactly as much memory as a float would, and it will be as efficient as a float. (The entire struct will be passed around in a register just as a float would.)
You can overload operators, including type cast operators, including implicit type cast operators, so if your temperature needs to be freely interchangeable with some other kind of entity, you can overload the appropriate operators and achieve this without any type casts. (See https://msdn.microsoft.com/en-us/library/85w54y0a.aspx)
It should be noted, however, that free interchangeability between data types defeats the purpose of declaring them as separate data types in the first place. What you should do instead is encapsulate as many operations as possible within the type itself, so as to minimize the situations where the type needs to be converted to something else.
Amendment
The practice of defining a special data type for every different type of quantity you are dealing with is derived from, and in accordance with, the practice of Physicists to always follow every single term of an equation with the units of that term. So, length is never just a plain 5 value, it is always a 5 metres value (in SI), and also acceleration is never just a plain 9.8 value, it is always a 9.8 metres per second squared value. This tends to make it very evident when a mistake is being made, and therefore minimizes the chances of it happening. For example, attempting to use a value in centimeters in an equation which expects meters, and also (even more importantly) attempting to use a velocity value in an equation which expects an acceleration value. For more information about this, see, for example, this: https://physics.stackexchange.com/q/138841
Making mistakes evident was also the original intention of Hungarian notation (see https://en.wikipedia.org/wiki/Hungarian_notation) though at the time it was not really possible with the C language to define a different data type for each kind of value and maintain performance. So, they just fiddled with the names of the variables.
With modern languages like C# we can finally declare our own efficient types, so that data type substitution mistakes are not just easier to catch, but actually impossible to make.

Before looking at the performance, see if the correctness of your program is compromised by this assignment:
one = (int)two;
it has a potential of overflowing int, which is much worse than a simple loss of a few CPU cycles.
If you know that there will be no overflow, because you have sanitized the values of unsigned variables (in this case, two is less than or equal to int.MaxValue) then you might as well use int. The reasons to go for an unsigned data type would be to gain the extra bit, or to make all operations treat the number as unsigned, not to gain performance or to make an implicit validation.

In .NET Framework, even for positive or zero values, Microsoft choosed to use int or long in place of uint or ulong. Why : because it's easier to avoid type casts between signed and unsigned types and potential exception on sign change on substractions that unsigned integers may imply.
So, what does using only posivite or zero values on an int mean: Using only 31 of the 32 bits. Not a problem : we have a lot of memory.
This may become a problem if you want to store binary data, in that case you will generaly use byte that is unsigned. Data size can also become important for very large data storage and in communications on slow medias. Some optimized algorithms will be written to minimize memory usage and will also use unsigned values.
What is the good practice : If you use values between 0 and 2 000 000 000, use an int, if you want to be easy to use, use ints or longs. If memory usage impact, media bandwith usage is crucial, data are of binary nature (images bits, sounds, ...) or algorithm efficiency requires use of uint or ulong, use them but be carefull during substractions : substracting a number higher than the first operand will throw an ArithmeticOverflowException.

Choosing the right data type vs casting

I am trying to learn to write good C# code and i want your advice.
If i have for example an unsigned number value that is smaller than 255 so this value fits in the byte type. But this value will not be used as byte but as an Int to set the combobox index so it will needs casting.
My question is how much do I have to worry about data types when writing good C# code? Do I need to declare this value as an Int to avoid casting?
Thank you for your time.

Casting from byte to int is safe since int is wider than byte, so no data loss is possible.
However, if you are defining a variable as byte but actually casting it to int all the time, it's better to just make it int, since you would only be confusing yourself and other readers of your code.

If you need it to be an int, use an int.
Data types are important, but so is the issue of code readability - having to cast all over the place reduces readability.

The answer is yes.
When you need to use not very big integer value you should avoid using/casting to any other types but int unless you have the needs in it. Like limited bandwidth where you are sending this data.

Data typing is fundamental to good coding - it lets compilers do a lot of checking for you and therefore helps eliminate/highlight bugs.
In your example, you could use the ushort to store your number, but unless you're writing a very memory sensitive application, you will be better off using an int as many library calls will need to convert the parameter back to an integer anyway. (this will make your code less readable due to constant type casting - and possibly less efficient too)

With the amount of memory currently available on our computers, memory is not an issue anymore when it's time to choose between byte or int.
I'd add that an int is much easier to deal with.

Design of your data types should not depend on way how they displayed by controls (or saved to data storage). If I see property of type byte then it tells me what range of values my data type expects. E.g. Color class has self-descriptive properties R,G,B of type byte.
Also do not do premature optimizations - if it become performance problem (I have doubts that casting from byte to int could have big influence on system performance) then you can optimize problem code later.
BTW casting from byte to int is implicit, so you don't need to do it manually:
comboBox.SelectedIndex = myObj.ByteProperty;

Thank you guys for your answers. I couldn't accept one answer because they complement each other so I decided to group them.
The CPU is designed to work efficiently with 32-bit values (linked question).
There is no need to do premature optimization.
It is not efficient to always cast data types.
Declaring a variable of certain type and always casting it reduces code readability.

Benefits of using enums over directly using integral types?

1) I’m aware of the following benefits:
they increase the level of abstraction since you immediately see what underlying integral values represent.
You can use them instead of magic numbers and by doing that making the code more understandable
They also restrict the values an enum variable can have and in doing so make the application safer, since programmers know which values are valid for variable, so I guess they sort of provide a type safety
Are there any other benefits they provide over directly using integral values?
2) Why do they use integrals as an underlying type and not string?
thank you

You've listed a lot of the core reasons where enums are preferable to integral types.
Named constants are safer and more readable than magic numbers
Enums describe to programmers what they are for. Integral values don't.
Naturally limiting the set of values that can be passed in. (You've got the tip of the type-safety iceberg... but look deeper...)
You can also add:
Vastly increased Type Safety. If you accept an 'int', then any int can be passed in. If you accept a VehicleType, then only a VehicleType can be passed in. I'm not just talking about someone passing in 6 when the largest allowed number is 5. I mean what if you pass in FuelType.Unleaded to a function that thinks it means VehicleType.Aeroplane? With enums the compiler will tell you you're an idiot. An integral type says "yeah, 5 is fine with me" and your program exhibits really odd behaviour that may be extremely difficult to trace.
Easier refactoring. Just as with any magic constants, If you pass in the value 5 in a hundred places in your program, you're in trouble if you decide to change 5 to have a different meaning. With an enum (as long as you don't have binary backwards compatibility concerns) you can change the underlying values. You can also change the underlying type of an enum if you wish (byte -> int -> long) without having to do anything more than recompile the client code.
Bitfields are so much easier to work with when the bits and masks can be named. And if you add new bits, you can often arrange things so that merely updating the related masks will allow most of your existing code to handle the new bitfields perfectly without having to rewrite them from scratch.
Consistency throughout the program. If you are careful with obfuscation and type safety, enums allow you to represent a list of named values that a user chooses from with the same names in the code, but without the efficiency cost of using strings.
Everybody understands why constants are great in code. Enums simply give you a way of holding together a related group of constants. You could achieve the same thing in a messier manner using a namespace of consts.
Using an enum for a parameter rather than a bool not only makes the code self-documenting, readable, and less prone to mistakes. It also makes it much easier to add a third option when you realize that two options isn't enough.
As with all tools, enums can be misused. Just use them where they make sense.
2) Why use bytes or ints instead of strings? Simply they're small and efficient.

I would conjecture that they require underlying integral types to ensure simplicity of comparison and more easily support bit flags. Without that limitation, we, or the compiler, or the runtime, would likely have to resort to some fuzziness to do things like combinations - or we would get into a situation where - as you say - we shouldn't care about the underlying type (the point of the abstraction) and yet when we try to say A | B we get a runtime error because we used an underlying type that isn't capable of that type of operation.

One benefit is when you want to use enum as a flag.
So if you define an enum like this:
[Flags]
public enum TestEnum{ A, B, C, D };
Then if you have a method that accept an instance of TestEnum as a variable, you can combine the values from the enum, so you can send for example A | B | C as the parameter for the method. Then, inside the method, you can check the parameter like this:
if ((paramname & TestEnum.A) > 0)
{
//do things related to A
}
if ((paramname & TestEnum.B) > 0)
{
//do things related to B
}
//same for C and D
Also, I think the reasons you mention are good enough by themselves to use enums.
Also regarding the comment that you can force an wrong value into an enum with code like this (TestEnum)500; it's hard to do if you do not want to break your code.
The point that the value 0 for an enum should be the default value, or in the case of flags "the absence of all other flags" is very important, since the line TestEnum myenum will instanciate myenum as 0 regardless if you have defined any enum value for 0 or not.

You can also parse an Enum from the string representation. You may get that string from a data source or user-entry.
I think you sold me on Enums at "magic numbers".

The main benefit of enum is that constants can be referred to in a consistent, expressive and type safe way.
Readability is of-course the topmost advantage of using the enumeration.
Another advantage is that enumerated constants are generated automatically by the compiler.
For instance, if you had an enumerated constant type for error codes that could occur in your program, your enum definition could look something like this:
enum Error_Code
{
OUT_OF_MEMORY,
FILE_NOT_FOUND
};
OUT_OF_MEMORY is automatically assigned the value of 0 (zero) by the compiler
because it appears first in the definition.FILE_NOT_FOUND equal to 1, so on.
If you were to approach the same example by using symbolic constants or Magic numbers, you write much more code to do the same.

Why don't Java, C# and C++ have ranges?

Ada, Pascal and many other languages support ranges, a way to subtype integers.
A range is a signed integer value which ranges from a value (first) to another (last).
It's easy to implement a class that does the same in OOP but I think that supporting the feature natively could let the compiler to do additional static checks.
I know that it's impossible to verify statically that a variabile defined in a range is not going to "overflow" runtime, i.e. due to bad input, but I think that something could be done.
I think about the Design by Contract approach (Eiffel) and the Spec# ( C# Contracts ), that give a more general solution.
Is there a simpler solution that checks, at least, static out-of-bound assignment at compile time in C++, C# and Java? Some kind of static-assert?
edit: I understand that "ranges" can be used for different purpose:
iterators
enumerators
integer subtype
I would focus on the latter, because the formers are easily mappable on C* language .
I think about a closed set of values, something like the music volume, i.e. a range that goes from 1 up to 100. I would like to increment or decrement it by a value. I would like to have a compile error in case of static overflow, something like:
volume=rangeInt(0,100);
volume=101; // compile error!
volume=getIntFromInput(); // possible runtime exception
Thanks.

Subrange types are not actually very useful in practice. We do not often allocate fixed length arrays, and there is also no reason for fixed sized integers. Usually where we do see fixed sized arrays they are acting as an enumeration, and we have a better (although "heavier") solution to that.
Subrange types also complicate the type system. It would be much more useful to bring in constraints between variables than to fixed constants.
(Obligatory mention that integers should be arbitrary size in any sensible language.)

Ranges are most useful when you can do something over that range, concisely. That means closures. For Java and C++ at least, a range type would be annoying compared to an iterator because you'd need to define an inner class to define what you're going to do over that range.

Java has had an assert keyword since version 1.4. If you're doing programming by contract, you're free to use those to check proper assignment. And any mutable attribute inside an object that should fall within a certain range should be checked prior to being set. You can also throw an IllegalArgumentException.
Why no range type? My guess is that the original designers didn't see one in C++ and didn't consider it as important as the other features they were trying to get right.

For C++, a lib for constrained values variables is currently being implemented and will be proposed in the boost libraries : http://student.agh.edu.pl/~kawulak/constrained_value/index.html

Pascal (and also Delphi) uses a subrange type but it is limited to ordinal types (integer, char and even boolean).
It is primarilly an integer with extra type checking. You can fake that in an other language using a class. This gives the advantage that you can apply more complex ranges.

I would add to Tom Hawtin response (to which I agree) that, for C++, the existence of ranges would not imply they would be checked - if you want to be consistent to the general language behavior - as array accesses, for instance, are also not range-checked anyway.
For C# and Java, I believe the decision was based on performance - to check ranges would impose a burden and complicate the compiler.
Notice that ranges are mainly useful during the debugging phase - a range violation should never occur in production code (theoretically). So range checks are better to be implemented not inside the language itself, but in pre- and post- conditions, which can (should) be stripped out when producing the release build.

This is an old question, but just wanted to update it. Java doesn't have ranges per-se, but if you really want the function you can use Commons Lang which has a number of range classes including IntRange:
IntRange ir = new IntRange(1, 10);
Bizarrely, this doesn't exist in Commons Math. I kind of agree with the accepted answer in part, but I don't believe ranges are useless, particularly in test cases.

C++ allows you to implement such types through templates, and I think there are a few libraries available doing this already. However, I think in most cases, the benefit is too small to justify the added complexity and compilation speed penalty.
As for static assert, it already exists.
Boost has a BOOST_STATIC_ASSERT, and on Windows, I think Microsoft's ATL library defines a similar one.
boost::type_traits and boost::mpl are probably your best friends in implementing something like this.

The flexibility to roll your own is better than having it built into the language. What if you want saturating arithmetic for example, instead of throwing an exception for out of range values? I.e.
MyRange<0,100> volume = 99;
volume += 10; // results in volume==100

In C# you can do this:
foreach(int i in System.Linq.Enumerable.Range(0, 10))
{
// Do something
}

JSR-305 provides some support for ranges but I don't know when if ever this will be part of Java.

Should I use int or Int32

In C#, int and Int32 are the same thing, but I've read a number of times that int is preferred over Int32 with no reason given. Is there a reason, and should I care?

The two are indeed synonymous; int will be a little more familiar looking, Int32 makes the 32-bitness more explicit to those reading your code. I would be inclined to use int where I just need 'an integer', Int32 where the size is important (cryptographic code, structures) so future maintainers will know it's safe to enlarge an int if appropriate, but should take care changing Int32s in the same way.
The resulting code will be identical: the difference is purely one of readability or code appearance.

ECMA-334:2006 C# Language Specification (p18):
Each of the predefined types is shorthand for a system-provided type. For example, the keyword int refers to the struct System.Int32. As a matter of style, use of the keyword is favoured over use of the complete system type name.

They both declare 32 bit integers, and as other posters stated, which one you use is mostly a matter of syntactic style. However they don't always behave the same way. For instance, the C# compiler won't allow this:
public enum MyEnum : Int32
{
member1 = 0
}
but it will allow this:
public enum MyEnum : int
{
member1 = 0
}
Go figure.

I always use the system types - e.g., Int32 instead of int. I adopted this practice after reading Applied .NET Framework Programming - author Jeffrey Richter makes a good case for using the full type names. Here are the two points that stuck with me:
Type names can vary between .NET languages. For example, in C#, long maps to System.Int64 while in C++ with managed extensions, long maps to Int32. Since languages can be mixed-and-matched while using .NET, you can be sure that using the explicit class name will always be clearer, no matter the reader's preferred language.
Many framework methods have type names as part of their method names:
BinaryReader br = new BinaryReader( /* ... */ );
float val = br.ReadSingle(); // OK, but it looks a little odd...
Single val = br.ReadSingle(); // OK, and is easier to read

int is a C# keyword and is unambiguous.
Most of the time it doesn't matter but two things that go against Int32:
You need to have a "using System;" statement. using "int" requires no using statement.
It is possible to define your own class called Int32 (which would be silly and confusing). int always means int.

As already stated, int = Int32. To be safe, be sure to always use int.MinValue/int.MaxValue when implementing anything that cares about the data type boundaries. Suppose .NET decided that int would now be Int64, your code would be less dependent on the bounds.

Byte size for types is not too interesting when you only have to deal with a single language (and for code which you don't have to remind yourself about math overflows). The part that becomes interesting is when you bridge between one language to another, C# to COM object, etc., or you're doing some bit-shifting or masking and you need to remind yourself (and your code-review co-wokers) of the size of the data.
In practice, I usually use Int32 just to remind myself what size they are because I do write managed C++ (to bridge to C# for example) as well as unmanaged/native C++.
Long as you probably know, in C# is 64-bits, but in native C++, it ends up as 32-bits, or char is unicode/16-bits while in C++ it is 8-bits. But how do we know this? The answer is, because we've looked it up in the manual and it said so.
With time and experiences, you will start to be more type-conscientious when you do write codes to bridge between C# and other languages (some readers here are thinking "why would you?"), but IMHO I believe it is a better practice because I cannot remember what I've coded last week (or I don't have to specify in my API document that "this parameter is 32-bits integer").
In F# (although I've never used it), they define int, int32, and nativeint. The same question should rise, "which one do I use?". As others has mentioned, in most cases, it should not matter (should be transparent). But I for one would choose int32 and uint32 just to remove the ambiguities.
I guess it would just depend on what applications you are coding, who's using it, what coding practices you and your team follows, etc. to justify when to use Int32.
Addendum:
Incidentally, since I've answered this question few years ago, I've started using both F# and Rust. F#, it's all about type-inferences, and bridging/InterOp'ing between C# and F#, the native types matches, so no concern; I've rarely had to explicitly define types in F# (it's almost a sin if you don't use type-inferences). In Rust, they completely have removed such ambiguities and you'd have to use i32 vs u32; all in all, reducing ambiguities helps reduce bugs.

There is no difference between int and Int32, but as int is a language keyword many people prefer it stylistically (just as with string vs String).

In my experience it's been a convention thing. I'm not aware of any technical reason to use int over Int32, but it's:
Quicker to type.
More familiar to the typical C# developer.
A different color in the default visual studio syntax highlighting.
I'm especially fond of that last one. :)

I always use the aliased types (int, string, etc.) when defining a variable and use the real name when accessing a static method:
int x, y;
...
String.Format ("{0}x{1}", x, y);
It just seems ugly to see something like int.TryParse(). There's no other reason I do this other than style.

Though they are (mostly) identical (see below for the one [bug] difference), you definitely should care and you should use Int32.
The name for a 16-bit integer is Int16. For a 64 bit integer it's Int64, and for a 32-bit integer the intuitive choice is: int or Int32?
The question of the size of a variable of type Int16, Int32, or Int64 is self-referencing, but the question of the size of a variable of type int is a perfectly valid question and questions, no matter how trivial, are distracting, lead to confusion, waste time, hinder discussion, etc. (the fact this question exists proves the point).
Using Int32 promotes that the developer is conscious of their choice of type. How big is an int again? Oh yeah, 32. The likelihood that the size of the type will actually be considered is greater when the size is included in the name. Using Int32 also promotes knowledge of the other choices. When people aren't forced to at least recognize there are alternatives it become far too easy for int to become "THE integer type".
The class within the framework intended to interact with 32-bit integers is named Int32. Once again, which is: more intuitive, less confusing, lacks an (unnecessary) translation (not a translation in the system, but in the mind of the developer), etc. int lMax = Int32.MaxValue or Int32 lMax = Int32.MaxValue?
int isn't a keyword in all .NET languages.
Although there are arguments why it's not likely to ever change, int may not always be an Int32.
The drawbacks are two extra characters to type and [bug].
This won't compile
public enum MyEnum : Int32
{
AEnum = 0
}
But this will:
public enum MyEnum : int
{
AEnum = 0
}

I know that the best practice is to use int, and all MSDN code uses int. However, there's not a reason beyond standardisation and consistency as far as I know.

You shouldn't care. You should use int most of the time. It will help the porting of your program to a wider architecture in the future (currently int is an alias to System.Int32 but that could change). Only when the bit width of the variable matters (for instance: to control the layout in memory of a struct) you should use int32 and others (with the associated "using System;").

int is the C# language's shortcut for System.Int32
Whilst this does mean that Microsoft could change this mapping, a post on FogCreek's discussions stated [source]
"On the 64 bit issue -- Microsoft is indeed working on a 64-bit version of the .NET Framework but I'm pretty sure int will NOT map to 64 bit on that system.
Reasons:
1. The C# ECMA standard specifically says that int is 32 bit and long is 64 bit.
2. Microsoft introduced additional properties & methods in Framework version 1.1 that return long values instead of int values, such as Array.GetLongLength in addition to Array.GetLength.
So I think it's safe to say that all built-in C# types will keep their current mapping."

int is the same as System.Int32 and when compiled it will turn into the same thing in CIL.
We use int by convention in C# since C# wants to look like C and C++ (and Java) and that is what we use there...
BTW, I do end up using System.Int32 when declaring imports of various Windows API functions. I am not sure if this is a defined convention or not, but it reminds me that I am going to an external DLL...

Once upon a time, the int datatype was pegged to the register size of the machine targeted by the compiler. So, for example, a compiler for a 16-bit system would use a 16-bit integer.
However, we thankfully don't see much 16-bit any more, and when 64-bit started to get popular people were more concerned with making it compatible with older software and 32-bit had been around so long that for most compilers an int is just assumed to be 32 bits.

I'd recommend using Microsoft's StyleCop.
It is like FxCop, but for style-related issues. The default configuration matches Microsoft's internal style guides, but it can be customised for your project.
It can take a bit to get used to, but it definitely makes your code nicer.
You can include it in your build process to automatically check for violations.

It makes no difference in practice and in time you will adopt your own convention. I tend to use the keyword when assigning a type, and the class version when using static methods and such:
int total = Int32.Parse("1009");

int and Int32 is the same. int is an alias for Int32.

You should not care. If size is a concern I would use byte, short, int, then long. The only reason you would use an int larger than int32 is if you need a number higher than 2147483647 or lower than -2147483648.
Other than that I wouldn't care, there are plenty of other items to be concerned with.

int is an alias for System.Int32, as defined in this table:
Built-In Types Table (C# Reference)

I use int in the event that Microsoft changes the default implementation for an integer to some new fangled version (let's call it Int32b).
Microsoft can then change the int alias to Int32b, and I don't have to change any of my code to take advantage of their new (and hopefully improved) integer implementation.
The same goes for any of the type keywords.

You should not care in most programming languages, unless you need to write very specific mathematical functions, or code optimized for one specific architecture... Just make sure the size of the type is enough for you (use something bigger than an Int if you know you'll need more than 32-bits for example)

It doesn't matter. int is the language keyword and Int32 its actual system type.
See also my answer here to a related question.

Use of Int or Int32 are the same Int is just sugar to simplify the code for the reader.
Use the Nullable variant Int? or Int32? when you work with databases on fields containing null. That will save you from a lot of runtime issues.

Some compilers have different sizes for int on different platforms (not C# specific)
Some coding standards (MISRA C) requires that all types used are size specified (i.e. Int32 and not int).
It is also good to specify prefixes for different type variables (e.g. b for 8 bit byte, w for 16 bit word, and l for 32 bit long word => Int32 lMyVariable)
You should care because it makes your code more portable and more maintainable.
Portable may not be applicable to C# if you are always going to use C# and the C# specification will never change in this regard.
Maintainable ihmo will always be applicable, because the person maintaining your code may not be aware of this particular C# specification, and miss a bug were the int occasionaly becomes more than 2147483647.
In a simple for-loop that counts for example the months of the year, you won't care, but when you use the variable in a context where it could possibly owerflow, you should care.
You should also care if you are going to do bit-wise operations on it.

Using the Int32 type requires a namespace reference to System, or fully qualifying (System.Int32). I tend toward int, because it doesn't require a namespace import, therefore reducing the chance of namespace collision in some cases. When compiled to IL, there is no difference between the two.

According to the Immediate Window in Visual Studio 2012 Int32 is int, Int64 is long. Here is the output:
sizeof(int)
4
sizeof(Int32)
4
sizeof(Int64)
8
Int32
int
base {System.ValueType}: System.ValueType
MaxValue: 2147483647
MinValue: -2147483648
Int64
long
base {System.ValueType}: System.ValueType
MaxValue: 9223372036854775807
MinValue: -9223372036854775808
int
int
base {System.ValueType}: System.ValueType
MaxValue: 2147483647
MinValue: -2147483648

Also consider Int16. If you need to store an Integer in memory in your application and you are concerned about the amount of memory used, then you could go with Int16 since it uses less memeory and has a smaller min/max range than Int32 (which is what int is.)

It's 2021 and I've read all answers. Most says it's basically the same (it's an alias), or, it depends on "what you like", or "by convention use int..." No answer gives you a clear when, where and why use Int32 over int. That's why I'm here.
98% of the time, you can get away with int, and that's perfectly fine. What are the other 2% ?
IO with records (struct, native types, organization and compression). Someone said an useless application is one that can read and manipulate data, but not actually capable of writing new datas to a defined storage. But in order to not reinvent the wheel, at some point, those dealing with old datas has to retrieve the documentation on how to read them. And chances are they were compiled from an era where a long was always a 32-bits integer.
It happenned before, where some had trouble remembering a db is a byte, a dw is a word, a dd is a double word, but how many bits was that about ? And that will likely happen again on C# 43.0 on a 256-bits platform... where the (future) boys never heard of "by convention, use int instead of Int32". That's the 2% where Int32 matters over int. MSDN saying today it's recommended to use int is irrelevant, it usually works with current C# version, but that may get dropped in future MSDN pages, in 2028, or 2034 ? Fewer and fewer people have WORD and DWORD encouters today, yet, two decades ago, they were common. The same thing will happen to int, in the very case of dealing with precise-fixed-length data.
In memory, a ushort (UInt16) can be a Decimal as long as it's fractional part is null, it is positive or null, and does not exceed 65535. But inside a file, it must be a short, 16-bits long. And when you read a documentation about a file structure from another era (inside the source code), you realize there are 3545 records definitions, some nested inside others, each record having between a couple and hundreds of fields of varying types.
Somewhere in 2028 a boy thought he could just get away by Ctrl-H-ing int to Int32, whole word only and match case... ~67000 changes in whole solution. Hit Run and still get CTDs. Clap clap clap. Go figure which int you should have changed to Int32 and which ones you should have changed to var. Also worth to point out Pointers are useful, when you deal with terabytes of datas (have a virtual representation of an entire planet on some cloud, download on demand, and render to user screen). Pointers are really fast in the ~1% of cases where there are so many datas to compute in realtime, you must trade with unsafe code. Again, it's to come up with an actually useful application, instead of being fancy and waste time porting to managed. So, be carefull, IntPtr is 32-bits or 64-bits already ? Could you get away with your code without caring how many bytes you read/skip ? Or just go (Int32*) int32Ptr = (Int32*) int64Ptr;...
An even more factual example is a file containing data processing and their respective commands (methods in the source code), like internal branching (a conditional continue or jump to if the test fails) :
IfTest record in file says : if value equals someConstant, jump to address. Where address is a 16-bits integer representing a relative pointer inside the file (you can go back towards the start of the file up to 32768 bytes, or up to 32767 bytes further down). But 10 years later, platforms can handle larger files and larger datas, and now you have 32-bits relative address. Your method in the source code were named IfTestMethod(...), now how would you name the new one ? IfTestMethodInt() or IfTestMethod32() ? Would you also rename the old method IfTestMethodShort() or IfTestMethod16() ? Then a decade later, you get a new command with long (Int64) relative address... What about a 128 bits command some 10 years later ? Be consistent ! Principles are great, but sometimes logic is better.
The problem is not me or you writing a code today, and it appears okay to us. It is being in the place of the one guy trying to understand what we wrote, 10 or 20 years later, how much it costs in time (= money) to come up with a working updated code ? Being explicit or writing redundant comments will actually save time. Which one you prefer ? Int32 val; or var val; // 32-bits.
Also, working with foreign data from other platforms or compile directives is a concept (today involves Interop, COM, PInvoke...) And that's a concept we cannot get rid of, whatever the era, because it takes time to update (reformat) datas (via serialization for ex.) Upgrading DLLs to managed code also takes time. We took time to leave assembler behind and go full-C. We are taking time to move from 32-bits datas to 64-bits, yet, we still need to care about 8 and 16-bits. What next in the future ? Move from 128-bits to 256 or directly to 1024 ? Do not assume a keyword explicit to you will remain explicit for the guys reading your documentation 20 years later (and documentation usually contains errors, mainly because of copy/paste).
So here it is : Where to use Int32 today over int ?
It's when you are producing code that is data-size sensible (IO, network, cross-platform data...), and at some point in the future - could be decades later - someone will have to understand and port your code. The key reason is era-based. 1000 lines of code, it's okay to use int, 100000 lines, it's not anymore. That's a rare duty only a few will have to do, and hell yeah, they have struggle, if only some were a little more explicit instead of relying on "by convention" or "it looks pretty in the IDE, Int32 is so ugly" or "they are the same, don't bother, it's a waste of time to write that two numbers and holding shift key", or "int is unambiguous", or "those who don't like int are just VB fanboys - go learn C# you noob" (yeah, that's the underlying meaning of a few comments right here)
Do not take what I wrote as a generalized perception, nor an attempt to promote Int32 on all cases. I clearly stated the specific case (as it seems to me this was not clear from other answers), to advocate for the few ones getting blammed by their supervisors for being fancy writing Int32, and at the same time the very same supervisor not understanding what takes so long to rewrite that C DLL to C#. It's an edge case, but at least for those reading, "Int32" has at least one purpose in its life.
The point can be further discussed by turning the question the other way around : Why not just get rid of Int32, Int64 and all the other variants in future C# specifications ? What that would imply ?