Refactored the `IQuantity` interface #1587

lipchev · 2025-07-27T14:22:14Z

refactored the IQuantity interface: removed the Equals with tolerance and the ToString overloads
introduced IAffineQuantity (see Temperature), ILogarithmicQuantity (see AmplitudeRatio) and ILinearQuantity (see Mass)
introduced extension methods for Equals with tolerance as well as Sum and Average (as applicable)
replaced the implementation of the GetHashCode method with a call to a better optimized function in the Comparison class
removed the Sum/Average from UnitMath (replaced be the new extension methods using the more concrete definitions)
marked the GenericMathExtensions class as [Obsolete] (replaced by the new extension methods)

…ce and the ToString overloads - introduced IAffineQuantity (see Temperature), ILogarithmicQuantity (see AmplitudeRatio) and ILinearQuantity (see Mass) - introduced extension methods for Equals with tolerance as well as Sum/Average

lipchev · 2025-07-27T14:36:44Z

@angularsen This is the most minimalist version of the interface, applicable to the current code-base (which relies on the DefaultUnitConverter):

In the 🐲 PR there isn't the need for the (newly added) double As(UnitKey unitKey) method to exist on the interface (or the extending type), currently it's just there as a work-around the inability to use UnitConverter.Default.ConvertValue(quantity, toUnit) from the extension.
I've removed the very generic Equals(IQuantity, IQuantity) (as we have agreed in an earlier discussion)
I've also removed the concrete Equals with tolerance- it is now an extension method. It is possible to have it on the interface - just not on that level (as we need to be able to specify the type of tolerance):

This definition applies only to the ILinearQuantity and ILogarithmicQuantity

        /// <summary>
        ///     <para>
        ///     Compare equality to <paramref name="other"/> given a <paramref name="tolerance"/> for the maximum allowed +/- difference.
        ///     </para>
        ///     <example>
        ///     In this example, the two quantities will be equal if the value of b is within 0.01 of a (0.01m or 1cm).
        ///     <code>
        ///     var a = Length.FromMeters(2.0);
        ///     var b = Length.FromMeters(2.1);
        ///     var tolerance = Length.FromCentimeters(10);
        ///     a.Equals(b, tolerance); // true, 2m equals 2.1m +/- 0.1m
        ///     </code>
        ///     </example>
        ///     <para>
        ///     It is generally advised against specifying "zero" tolerance, due to the nature of floating-point operations.
        ///     </para>
        /// </summary>
        /// <param name="other">The other quantity to compare to.</param>
        /// <param name="tolerance">The absolute tolerance value. Must be greater than or equal to zero.</param>
        /// <returns>True if the absolute difference between the two values is not greater than the specified tolerance.</returns>
        bool Equals(TSelf? other, TSelf tolerance);

For the IAffineQuantity (e.g. the Temperature) we need to use this definition:

        bool Equals(TSelf? other, TOffset tolerance);

UnitsNet/ExceptionHelper.cs

UnitsNet/UnitMath.cs

angularsen

I didn't get to the end and have to run, but here are comments so far.

UnitsNet/UnitMath.cs

CodeGen/Generators/UnitsNetGen/QuantityGenerator.cs

CodeGen/Generators/UnitsNetGen/UnitTestBaseClassGenerator.cs

UnitsNet/UnitMath.cs

angularsen · 2025-07-27T17:52:04Z

UnitsNet/Extensions/AffineQuantityExtensions.cs

+        }
+
+        // // TODO see about this (I think the exception should take precedence to the null check, but the QuantityTests disagree)
+        // if (tolerance is not TOffset toleranceQuantity)


I agree, argument validation should happen first, so this should throw.

Oh shit, I had completely forgotten about this thing..

Note that originally we had the very generic Equals(IQuantity? other, IQuantity tolerance) and the more specific Equals(TSelf? other, TSelf tolerance)

Do we want to have an extension of type Equals<TQuantity>(this TQuantity quantity, IQuantity? other, IQuantity tolerance) with a type check on both the other and the tolerance, with just the second one throwing (kinda weird IMO)

Or would we rather have Equals<TQuantity>(this TQuantity quantity, IQuantity? other, TQuantity tolerance) (with TOther tolerance for the affine quantities): no exceptions thrown (returns false when other is not null or another type)

Do we want these at all?

No, that's weird to me too. I say remove it.

Yes, I think those look useful enough. I assume you meant TOffset tolerance for affine quantities.

I think nr 2 is useful, unless we are already covered elsewhere that I'm not remembering?

It seems a bit redundant to have EqualsAbsolute(), as nr 1 and nr 2 just delegate to it. As far as I can see, the only difference is nullability. Can it simply be inlined? Or just renamed to Equals() as the absolute part in the name makes me think it serves a special purpose.

I'm not sure I understand- Equals(TSelf? other, TSelf tolerance) this overload is a given, my 1) 2) 3) were alternatives for the very generic overload Equals(IQuantity? other, IQuantity tolerance). Do you mean you pick option 2) as in (TQuantity, IQuantity, TQuantity) for the non-affine and (TQuantity, IQuantity, TOffset) for the temperature (which is already the case).

Regarding 4) - so you mean you'd like me to make the EqualsAbsolute internal (it's being used by the two other overloads)?

Okay I misread, IQuantity and TQuantity reads very similar.

For tolerance equality, I'd generally expect the fully typed TQuantity experience, for both other and tolerance to match this. And TOffset for affine.

But I guess, why not also allow untyped IQuantity and do type checking internally. It may be useful in certain cases and it costs us little to add. The only thing I'm unsure of, is whether it can trip up intellisense, auto-complete and AI coding assistance to not prefer the typed TQuantity.

Rename EqualsAbsolute to Equals, if it works without issues, or make it private if it's only intended to be used by the other extension methods.

fixed in bdf04d5 - please double-check the signatures of the Equals extensions.

Ref nr 2, see comment:
bdf04d5#r163347229

The link doesn't work, but I read it in the notification:

public static bool Equals(this IQuantity quantity, IQuantity? other, IQuantity tolerance)

This is only going to work if only one of the extensions is visible- (IQuantity, IQuantity, IQuantity) would be ambiguous with (TQuantity, TQuantity, TQuantity)..

UnitsNet/Extensions/QuantityExtensions.cs

UnitsNet/GeneratedCode/Quantities/Angle.g.cs

angularsen

Final batch of comments

UnitsNet/GeneratedCode/Quantities/Angle.g.cs

UnitsNet/GeneratedCode/Quantities/Temperature.g.cs

angularsen · 2025-07-27T19:45:52Z

UnitsNet/IArithmeticQuantity.cs

 /// </summary>
 /// <typeparam name="TSelf">The type itself, for the CRT pattern.</typeparam>
 /// <typeparam name="TUnitType">The underlying unit enum type.</typeparam>
-public interface IArithmeticQuantity<TSelf, TUnitType> : IQuantity<TSelf, TUnitType>
+public interface IArithmeticQuantity<TSelf, TUnitType> : IQuantity<TSelf, TUnitType>, ILinearQuantity<TSelf>


I think it'll be more to clear to name it ILinearArithmeticQuantity, if it's arithmetic is for linear quantities.

Do you mean to rename the IArithmeticQuantity interface (which would be a breaking change) or the ILinearQuantity interface?

Gotcha, no I meant renaming IArithmeticQuantity to ILinearArithmeticQuantity since it's now scoped to only linear quantities, where as before it covered all quantities.

Or is it just that we haven't implemented for affine/logarithmic yet, but this may come later?
We can discuss this more later, not important.

UnitsNet/IQuantity.cs

…arison class - Temperature: removed the [Obsolete] methods - added the missing tests for the Linear/Affine/Logarithmic extensions

angularsen · 2025-08-02T20:53:43Z

LGTM and ready to merge, if you want to address any of the open comments before you merge just do so.

As discussed in #1587, these are supposed to go away in `v6`

- replaced the type of the tolerance parameter of the Equals extension with the concrete quantity type

…evant usages from the benchmarks

lipchev · 2025-08-03T07:02:05Z

LGTM and ready to merge, if you want to address any of the open comments before you merge just do so.

I've addressed most of the issues, with the exception of the UnitMath for which I'll open a separate issue/PR. Please double-check the signatures of the Equals extensions and if everything looks good, then merge away..

lipchev · 2025-08-03T07:46:50Z

Not that anyone cares about size, but in case you're curious- here's the before and after:

Before (netstandard2.0):
2.46 MB (2 580 480 bytes)

After (netstandard2.0):
2.42 MB (2 544 128 bytes)

🐲 PR (nestandard2.0):
1.77 MB (1 864 192 bytes)

Before (net9.0):
2.55 MB (2 682 368 bytes)

After (net9.0):
2.49 MB (2 616 832 bytes)

🐲 PR (net9.0):
1.80 MB (1 888 256 bytes)

lipchev · 2025-08-03T16:33:38Z

@angularsen Actually there two more extensions that I could have used here:

    public static TQuantity ToUnit<TQuantity>(this TQuantity quantity, UnitSystem unitSystem)
        where TQuantity : IQuantityOfType<TQuantity>

and

    public static double As<TQuantity>(this TQuantity quantity, UnitSystem unitSystem)
        where TQuantity : IQuantity

tell me if you rather have them here, or in a separate PR.

In the 🐲 I have these in a separate file: UnitConversionExtensions, but I think they could simply go into QuantityExtensions

angularsen · 2025-08-03T18:01:01Z

tell me if you rather have them here, or in a separate PR.

Separate PR, let's get this bad boy merged. Easier to discuss smaller changes.
Regarding Equals, I have one suggestion #1587 (comment), but we can do that in separate PR also.

angularsen · 2025-08-03T18:02:05Z

I'll just merge this for now.

fixed the json-converter usage of Equals with (generic) tolerance

00daab2

lipchev commented Jul 27, 2025

View reviewed changes

UnitsNet/ExceptionHelper.cs Show resolved Hide resolved

lipchev commented Jul 27, 2025

View reviewed changes

UnitsNet/UnitMath.cs Show resolved Hide resolved

Temperature.extra.cs: Typo

5a4b8b7

angularsen mentioned this pull request Jul 27, 2025

Calculations with db are wrong #1569

Open

angularsen reviewed Jul 27, 2025

View reviewed changes

angularsen approved these changes Jul 27, 2025

View reviewed changes

lipchev added 2 commits July 29, 2025 03:23

- replaced the GetHashCode implementation with a helper from the Comp…

2ea2f0b

…arison class - Temperature: removed the [Obsolete] methods - added the missing tests for the Linear/Affine/Logarithmic extensions

merged the typo fix

9203e95

lipchev added the breaking-change label Jul 29, 2025

lipchev added 2 commits July 30, 2025 19:09

Merge remote-tracking branch 'upstream/master' into iquantity-interfaces

481d74c

splitting the IQuantity interface definitions into separate files

f9976ba

lipchev mentioned this pull request Jul 30, 2025

Deprecating the Multiply/Divide methods for the Temperature #1589

Merged

angularsen pushed a commit that referenced this pull request Aug 2, 2025

Deprecating the Multiply/Divide methods for the Temperature (#1589)

4b86df5

As discussed in #1587, these are supposed to go away in `v6`

lipchev added 3 commits August 3, 2025 09:15

- renamed IQuantityInstance to IQuantityOfType

bdf04d5

- replaced the type of the tolerance parameter of the Equals extension with the concrete quantity type

making the GenericMathExtensions class obsolete and replacing the rel…

a6fffc7

…evant usages from the benchmarks

remove the GenerateArithmetic from the Temperature.json

5d25e1d

This was referenced Aug 3, 2025

Making the methods from UnitMath that exist in System.Linq as [Obsolete] #1593

Merged

Replaced the NotImplementedException from ToUnit with UnitNotFoundException #1594

Merged

angularsen merged commit 5c49fb0 into angularsen:master Aug 3, 2025
1 check passed

angularsen mentioned this pull request Aug 9, 2025

Replacing the ToString overloads with extensions #1552

Closed

lipchev mentioned this pull request Aug 9, 2025

Replace As(UnitSystem) and ToUnit(UnitSystem) with extension methods #1600

Merged

Refactored the IQuantity interface #1587

Refactored the IQuantity interface #1587

Uh oh!

Conversation

lipchev commented Jul 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lipchev commented Jul 27, 2025

Uh oh!

Uh oh!

Uh oh!

angularsen left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

angularsen Jul 27, 2025

Choose a reason for hiding this comment

Uh oh!

lipchev Jul 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

angularsen Aug 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lipchev Aug 2, 2025

Choose a reason for hiding this comment

Uh oh!

angularsen Aug 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lipchev Aug 3, 2025

Choose a reason for hiding this comment

Uh oh!

angularsen Aug 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lipchev Aug 3, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

angularsen left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

angularsen Jul 27, 2025

Choose a reason for hiding this comment

Uh oh!

lipchev Jul 28, 2025

Choose a reason for hiding this comment

Uh oh!

angularsen Aug 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

angularsen commented Aug 2, 2025

Uh oh!

lipchev commented Aug 3, 2025

Uh oh!

lipchev commented Aug 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lipchev commented Aug 3, 2025

Uh oh!

angularsen commented Aug 3, 2025

Uh oh!

Refactored the `IQuantity` interface #1587

Refactored the `IQuantity` interface #1587

lipchev commented Jul 27, 2025 •

edited

Loading

lipchev Jul 28, 2025 •

edited

Loading

angularsen Aug 2, 2025 •

edited

Loading

angularsen Aug 2, 2025 •

edited

Loading

angularsen Aug 3, 2025 •

edited

Loading

angularsen Aug 2, 2025 •

edited

Loading

lipchev commented Aug 3, 2025 •

edited

Loading