openxmldev / linq-to-xml Goto Github PK

This looks like an interesting project.

Just a note, tsdotnet already has classes that seem to fit some of your use cases:

StringBuilder

https://github.com/tsdotnet/string-builder

LinqIterable

https://github.com/tsdotnet/linq/blob/master/src/linqExtended.ts
You might try inheriting from LinqExtended<T> instead of writing your own.

I love how your API is turning out.

I am still a bit concerned about making assumptions in your LinqIterableBase about what the defaults should be. In respect to selecting 'nodes', null is obviously a valid default, but you (or I) might be better served providing an .xOrNull() method to be more explicit.

// TypeScript
textNodes.select(parseFloat).firstOrDefault() // if textNodes is empty, the value returned here is null.

// C#
textNodes.Select(double.Parse).FirstOrDefault() // if textNodes is empty, the value returned here is 0.

As I pointed out before, when calling an .xOrDefault(), if for some reason T is of type number you won't get 0 as the default as you would in C#. And if you were attempting to filter by type: typeof null == 'object' where typeof undefined == 'undefined'. So the behavior differs even though you are conforming to the .NET API.

Where I see having null as an acceptable default is whenever the expected type should be an object like a node. It's just difficult to forsee how things might get weird if some amount of transformation is done by the consumer.

I'd also point to the weirdness in JS around 0 and null. Really strange things can happen that simply won't happen in C#.
https://www.tutorialspoint.com/what-is-the-relation-between-null-and-0-in-javascript
https://blog.campvanilla.com/javascript-the-curious-case-of-null-0-7b131644e274?gi=bc5bbbc977f4

null > 0 || null == 0; // == false
undefined > 0 || undefined == 0; // == false
undefined >= 0 // == false
null >= 0; // == true WHAT!?