Why couldn't this stay as

public DistributedApplicationExecutionContextOptions(DistributedApplicationOperation operation, string? publisherName)

Had a call with @mitchdenny and if you want to no set a valid publishname you can use the other ctor.

In general, that kind of API pattern makes using the API harder to use. It maybe doesn't matter much for this API, but in general, the pattern isn't great.

void Foo() { ... }
void Foo(string v) { ... }

// consume:

string? someV = GetSomeV();

if (someV is null)
{
    Foo();
}
else
{
    Foo(someV);
}

vs.

void Foo() { ... }
void Foo(string? v) { ... }

// consume:

string? someV = GetSomeV();
Foo(someV);

Got it. In that case the usage is based on the first argument, operation, and if you are creating a Publish operation you have to set a non-null publisherName. So maybe the API could have been more specific about it, even maybe preventing from setting a publisherName if you create another type (Run or Inspect). For instance ctor(string publisherName) would be more adapted, it would infer the Publish operation. Or maybe these should be sub-classes.