Bug with DateTime and leap year

[Tragen]

Description

If you add one year to 2024-02-29, the expected result differs from the actual result

Reproduction Steps

Create a console application

		DateTime d = new DateTime(2024, 02, 29);
		d = d.AddYears(1);
		Console.WriteLine(d.ToString());

Expected behavior

It should print the first of March 2025 and not the 28th of February 2025.
VB6, JavaScript and more languages do it correctly.

Actual behavior

.NET outputs 28th of February 2025.

Regression?

No response

Known Workarounds

No response

Configuration

Tested on .NET 4.8 until .NET 9 on Windows x64 and in https://dotnetfiddle.net/Hxy8gt

Other information

No response

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-datetime
See info in area-owners.md if you want to be subscribed.

[huoyaoyuan]

Then handling of leap date is very arbitrarily chosen and subjective. It also applies to AddMonth on month ends. There's no behavior that's more "correct", as long as they are consistent.

[Clockwork-Muse]

FWIW, Java also uses this "incorrect" behavior:

This method adds the specified amount to the years field in three steps:

 1. Add the input years to the year field
 2. Check if the resulting date would be invalid
 3. Adjust the day-of-month to the last valid day if necessary

For example, 2008-02-29 (leap year) plus one year would result in the invalid date 2009-02-29 (standard year). Instead of returning an invalid result, the last valid day of the month, 2009-02-28, is selected instead.

Note that, while the behavior is technically arbitrary, retaining the same month is actually beneficial for "striding" implementation use.

That aside, for this and similar month-to-month behavior, you should usually avoid end-of-month dates when possible, and I personally tend to regard any math on such values other than +/- days as concerning. You're usually better off working with start-of-month values.
If this fell out of real-world code, what was this value intended to represent?

[Clockwork-Muse]

As designed.
Even if we decided this was actually a bug, the change would be so massively breaking it would be rejected anyways.