Spike/idg modifications #36
Conversation
Convert to GitHub flavoured Markdown
Moved some stuff into placeholders in later chapters.
content/11_SchedulingAndThreading.md
Outdated
|
|
||
| Let's follow exactly what happens here. First, assume that this code is just running normally and not in any unusual context—perhaps inside the `Main` entry point of a program. When this code calls `Subscribe` on the `IObservable<int>` returned by `SelectMany`, that will in turn will call `Subscribe` on the `IObservable<int>` returned by the first `Observable.Range`, which will in turn schedule a work item for the generation of the first value in the range (`1`). | ||
|
|
||
| Since we didn't pass a scheduler explicitly to `Range`, it will use its default choice, the `CurrentThreadScheduler`, and that will ask itself "Am I already in the middle of handling some work item on this thread?" In this case the answer will be no, so it will run the work item immediately (before returning from the `Schedule` call made by the `Range` operator). The `Range` operator will then produce its first value, calling `OnNext` on the `IObserver<int>` that the `SelectMany` operator provided when it subscribed to the range. |
There was a problem hiding this comment.
will be no, so it will
The so feels out of place
There was a problem hiding this comment.
Really? The sentence makes no sense if I remove it. Something needs to signify that the remainder of the sentence is a consequence of the answer being no.
I've put the no in quotes, which I think makes its proximity to "so" look less odd.
| This uses an extension method that Rx supplies for cases where you have no need for the `state` argument. In fact it's the wrong thing to use here, because this could end up passing the wrong scheduler. If you look at the `IScheduler` interface, you'll see that the work item callback's first argument is an `IScheduler`, and you're supposed to use that scheduler for any logically recursive work. Some schedulers (e.g., the `ImmediateScheduler`) supply a special type of scheduler specifically to avoid logical recursion turning into actual recursion that could end up going so deep that the application crashes with a stack overflow. This particular extension method isn't meant for these logically recursive scenarios, so we can't quite make it this simple. But perhaps we could avoid that weirdness of passing our own `this` argument: | ||
|
|
||
| ```cs | ||
| var next = scheduler.Schedule<object?>(null, (innerScheduler, _) => LoopRec(innerScheduler)); |
There was a problem hiding this comment.
So is the this manifested as the discard?
There was a problem hiding this comment.
I don't understand the question. I'm not sure how to change the text to prevent a misunderstanding because I can't work out what you think is happening here.
The discarded parameter here is where the state gets passed back into our callback. In this case we've passed the Schedule method a state of null. So the thing being discarded here is null.
I've realised that one of the examples this section showed was a bit pointless, so I've removed it, and modified the discussion to fit, so I'm hoping that with that, it will be easier to follow what's going on.
There was a problem hiding this comment.
But perhaps we could avoid that weirdness of passing our own
thisargument:
I don't see "our own this argument" - where is it in the code???
There was a problem hiding this comment.
This is a reference to the preceding example, which looks like this:
var next = scheduler.Schedule(this, static (innerScheduler, @this) => @this.LoopRec(innerScheduler));I've tweaked the text again to make it clear that I'm referring to the earlier example.
| And some examples of some sources that might make good cold observables: | ||
|
|
||
| * the contents of a collection (such as is returned by the [`ToObservable` extension method for `IEnumerable<T>`](03_CreatingObservableSequences.md#from-ienumerablet)) | ||
| * a fixed range of values, such as [`Observable.Range`](03_CreatingObservableSequences.md#observablerange) produces |
There was a problem hiding this comment.
03_CreatingObservableSequences.md#observablerange
Should be 03_CreatingObservableSequences.md#observable.range
There was a problem hiding this comment.
Should it? VS code thinks it's right as it is. If I add the full stop as you suggest, the link is now broken inside VS Code.
Is this another "opinion is divided on the matter" markdownism?
There was a problem hiding this comment.
It might be two conflicting approaches.
The Id generated by that heading is https://www.introtorx.com/chapters/creating-observable-sequences#observable.create
There was a problem hiding this comment.
and is defined by markdig as "auto-identifiers" https://github.com/xoofx/markdig/blob/master/src/Markdig.Tests/Specs/AutoIdentifierSpecs.md
|
|
||
| * the contents of a collection (such as is returned by the [`ToObservable` extension method for `IEnumerable<T>`](03_CreatingObservableSequences.md#from-ienumerablet)) | ||
| * a fixed range of values, such as [`Observable.Range`](03_CreatingObservableSequences.md#observablerange) produces | ||
| * events generated based on an algorithm, such as [`Observable.Generate`](03_CreatingObservableSequences.md#observablegenerate) produces |
There was a problem hiding this comment.
03_CreatingObservableSequences.md#observablegenerate
Should be 03_CreatingObservableSequences.md#observable.generate
| * a fixed range of values, such as [`Observable.Range`](03_CreatingObservableSequences.md#observablerange) produces | ||
| * events generated based on an algorithm, such as [`Observable.Generate`](03_CreatingObservableSequences.md#observablegenerate) produces | ||
| * a factory for an asynchronous operation, such as [`FromAsync`](03_CreatingObservableSequences.md#from-task) returns | ||
| * events produced by running conventional code such as a loop; you can create such sources with [`Observable.Create`](03_CreatingObservableSequences.md#observablecreate) |
There was a problem hiding this comment.
03_CreatingObservableSequences.md#observablecreate
Should be 03_CreatingObservableSequences.md#observable.create
Inexplicably, this chapter was written mostly in HTML in a `.md` file, which confused some parts of our tooling. I've recast it all as markdown. I've also written a short note to say you can use most of these disposable helpers in non-Rx code too.
…ked Felix's suggestion
…erstood as I thought
First two chapters currently have more or less the structure and examples I want.