[xperiandri]

Proposed Changes

Builds on #35. SortedSet.exists and SortedSet.forall walk a set of a reference type through IEnumerable<T> instead of the struct enumerator of ImmutableSortedSet<T>, and a set of a value type as before. In the code that all reference types share, each call of a member of a generic struct needs a lookup of the exact type at run time, and the struct enumerator of ImmutableSortedSet<T> is made of such calls. No public signature changes.

	strings	integers
a walk through the interface against the struct enumerator, 100 to 100 000 elements	0.53 to 0.58 of the time (13 against 23 ns per element)	1.3 to 1.7
exists and forall of the library against a copy of the old loop, 100 to 100 000 elements, two runs	0.47 to 0.57	0.93 to 1.05, the same time

Ratios are the time relative to the code before; below 1 is faster. The cost of the interface is the boxed enumerator, 64 bytes per call, which the comment of the functions already counted for Seq.exists. The base build read 0.94 to 1.08 against the same copy of the old loop, which is the check of the script.

The other collections were measured for the same pattern too; only SortedSet and SortedMap gain (the method, and the tables of every collection and cell, are in the fifth comment of #45: #45 (comment)).

The contracts run on integers, so the commit adds tests with strings: the results of Set for the empty set and for the first, a middle, the last and an absent element, and the stop at the first match and at the first failure.

Types of changes

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)

A performance change that keeps every result and every signature.

Checklist

• Build and tests pass locally (the whole test project: 11 808 test runs on net8.0 and net10.0)
• I have added tests that prove my fix is effective or that my feature works
• I have added necessary documentation: the comment above exists and forall carries the measured numbers, as the other comments of the module do

Further comments

filter and partition also walk with the struct enumerator, but a builder adds every element they keep. They are unchanged and were not measured.

🤖 Generated with Claude Code