deno-puppeteer VS MediatR

In an ideal world, yes, but in reality Deno's permission model is quite hard to use so many libraries require disabling a lot of it, sometimes everything like puppeteer. Other example is the library I maintain: it requires --allow-net since you can't whitelist a domain and all its subdomains, just a domain.

import puppeteer from "https://deno.land/x/[email protected]/mod.ts"; const browser = await puppeteer.launch(); const page = await browser.newPage(); await page.goto("https://news.ycombinator.com", { waitUntil: "networkidle2", }); await page.pdf({ path: "hn.pdf", format: "A4" }); await browser.close();

I agree that semantics for that is complicated, but I think the adopting browser permission model is bad because browsers have per-site isolation but how Deno is going to do that for applications that require executing external binaries? Also, even if you would whitelist some binaries, there are Deno packages, such as deno-puppeteer, which don't list what permissions it requires, instead it instructs users to enable all permissions using -A. By the way, why just -A? Why not --unsafely-enable-all-permissions (like Chromium's --unsafely-treat-insecure-origin-as-secure)?

Maybe people forget about this permission system because either are not experienced with Deno or because they just slap -A on eveything. Some packages such as deno-puppeteer even put it in all examples without even adding a note about its risks.

You probably noticed that .SetName() returns a Either. You may have come across Unit in libraries like MediatR or Language-Ext. It's a simple construct representing a type with only one possible value. We use it as a placeholder for operations that do not return a value but may return another state. In our example, .SetName() is a Command that does not return a value but may fail. Therefore, the monad Either carries two possible states: Right (without value) or Left (with an Error).

The widely used MediatR library[0] could be used to do that as well, just FYI.

[0]: https://github.com/jbogard/MediatR

Hope this GH issue shed some light on why injecting handler inside another handler is not good https://github.com/jbogard/MediatR/issues/400

From: https://github.com/jbogard/MediatR

I gave it a go and I was impressed how easy and fast it was to set it all up. Since I'm not a big fan of REPR pattern almost all my projects are using CQRS pattern with a help of MediatR ](https://github.com/jbogard/MediatR) I immediately started going over something similar that Fast Endpoints offer which is a command bus.

I'd use a dedicated event bus based on Reactive Extensions or MediatR to publish domain events from your domain services. This probably doesn't solve all your ViewModel update problems as is, maybe you need to revise the granularity (maybe you can have smaller ViewModels that refresh single property that exposes the Model) and lifespan (sometimes you can create a ViewModel, make it perform it's task and then discard it completely) of your ViewModels.

If you're finding yourself in a situation where you need to turn this behavior into a pattern because there are a lot of View Models that need to execute async business logic in response to some changes, I'd go with something like MediatR or Reactive Extensions. The idea is, again, that some other, probably business-level, component listens to changes in a decoupled way (that means it doesn't subscribe directly to your View Model, but to an event bus instead). View Model publishes change events to the event bus, and business-component reacts to these events by executing the business logic.

What people really want is to design the logic of an app independently from the component hierarchy. That means you need to store state somewhere other than the components and you need to dispatch events that are not attached to the component hierarchy. Also, a one way data flow has well known benefits as described by things like CQRS, RabbitMQ, and MediatR.