Version Numbers float, decimal or double

Question

I have a document management system where documents can have multiple versions. Each version is saved and users can view version history.

What I would like to know is: What datatype should I use for version numbers? Decimal, Float or Double? I'm using .NET and C#.

Version numbers start at 0.1 and each published major version will be rounded to the next whole number. i.e. 0.4 goes to 1.0 and 1.3 goes to 2.0 etc.

When a version numbers hits 0.9 and a minor version is added I would like the number to go to 0.10 not 1.0, when I add to it. This is the biggest issue.

Any suggestions are appreciated.

Thanks.

Absorptance

Answer 1

System.Version

This already stores the different parts, deals with presenting it as a string (revision and build components are only used in display if they are non-zero, so their irrelevance to your case doesn't matter) and (best of all) is already understood by other .NET developers, and won't lead to confusion (if I saw some use of a version number that wasn't a System.Version I'd spend some time then trying to work out why Version wasn't good enough for the job, in case that proved important and hid a nasty surprise. If it was good enough for the job, I'd be irritated at the developer wasting my time like that).

You can deal with the means you want for incrementing easily with extension methods:

public static Version IncrementMajor(this Version ver)
{
  return new Version(ver.Major + 1, 0);
}
public static Version IncrementMinor(this Version ver)
{
  return new Version(ver.Major, ver.Minor + 1);
}

Kansas

Not sure why your solution is so low scored. In my opinion we should always use existing implementations when possible. – Nahuatlan

I would love to give t his an upvote, but I have used my quota for the day, so I have to come back and add an upvote tomorrow. Clearly the best answer in here. – Covington

At least good that the OP changed the checkmark from my answer to Jon Hanna's to his answer is shown at the top. – Covington

What ver.Minor is max value? – Barehanded

@Evgeny, do you mean what would happen if ver.Minor was set at int.MaxValue when you called IncrementMinor? You'd either get an overflow error or it would change to int.MinValue depending on the checked or unchecked context. This would make compiling the above checked (the default) to make sure the exception would happen, a good idea. – Kansas

Answer 2

How about two integers? One for major and one for minor revisions?

Rutherfordium

Answer 3

Make your own data type for this

public struct VersionNumber : IComparable<ReleaseNumber>
{
  public int MajorVersion{get;set;}
  public int MinorVersion{get;set;}

  public VersionNumber( int major, int minor)
  {
    MajorVersion = major;
    MinorVersion = minor;
  }

  public override string ToString(){
    return major + '.' + minor;
  }

  public int CompareTo(VersionNumber other) {
    int result;
    result = MajorVersion.CompareTo(other.MajorVersion);
    if (result != 0) { return result; }
    return MinorVersion.CompareTo(other.MinorVersion);
  }
  public static bool operator <(VersionNumber left, VersionNumber right) {
    return left.CompareTo(right) < 0;
  }
  public static bool operator <=(VersionNumber left, VersionNumber right) {
    return left.CompareTo(right) <= 0;
  }
  public static bool operator >(VersionNumber left, VersionNumber right) {
    return left.CompareTo(right) > 0;
  }
  public static bool operator >=(VersionNumber left, VersionNumber right) {
    return left.CompareTo(right) >= 0;
  }
}

You can also add a comparer so you can check two version numbers to see which one is the highest version of two version numbers for example.

EDIT

Added the comparer logic also for good measure :)

Scalade

Finally, I have a reason to use a struct. Thanks. – Absorptance

I wish I could +Jon Hanna's answer again. Don't reinvent the wheel – Slacken

I was unaware of the System.Version class, so I learned something here as well. i didn't intentionally reinvent the wheel. I just didn't know the wheel already existed :) – Covington

Answer 4

I would suggest two integers: a major and a minor. You can even store this as major * 1000 + minor if you want one variable.

Lindseylindsley

Answer 5

Decimal should be the best of the above given, but as other has noted two ints would be better.

Doubles and floats does not accurately store all decimal values, you don't want your version to suddenly be 1.219999999999999999

Comprehensive