Introduction

This post explains a simple pitfall where C# developers trying out F# may fail when writing async code.
Consider this simple code in which we download page contents using Puppeteer-sharp:
  1. let renderHtml = async {
  2. BrowserFetcher().DownloadAsync(BrowserFetcher.DefaultRevision) |> Async.AwaitTask |> ignore
  3. let options = LaunchOptions()
  4. options.Headless <- true
  5. let! browser = Puppeteer.LaunchAsync(options) |> Async.AwaitTask
  6. let! page = browser.NewPageAsync() |> Async.AwaitTask
  7. page.GoToAsync("https://i.ua") |> Async.AwaitTask |> ignore
  8. return! page.GetContentAsync() |> Async.AwaitTask
  9. }
Since we actually don't care about download browser result, we naturally would expect our line to look like:
  1. BrowserFetcher().DownloadAsync(BrowserFetcher.DefaultRevision) |> Async.AwaitTask |> ignore

This would be equivalent to the following C# code:

  1. await new BrowserFetcher().DownloadAsync(BrowserFetcher.DefaultRevision);
However, when we execute the following method, we get AggregateException which in turn contains the following inner exception: "Chromium revision is not downloaded. Run BrowserFetcher.DownloadAsync or download Chromium manually". It seems like we've called the following...
  1. let! browser = Puppeteer.LaunchAsync(options) |> Async.AwaitTask

...without waiting for the BrowserFetcher result.

And indeed, in order to await async call, we have to use let! as the construct. The code below works as expected:

  1. let renderHtml = async {
  2. let! _ = BrowserFetcher().DownloadAsync(BrowserFetcher.DefaultRevision) |> Async.AwaitTask
  3. let options = LaunchOptions()
  4. options.Headless <- true
  5. let! browser = Puppeteer.LaunchAsync(options) |> Async.AwaitTask
  6. let! page = browser.NewPageAsync() |> Async.AwaitTask
  7. let! _ = page.GoToAsync("https://i.ua") |> Async.AwaitTask
  8. return! page.GetContentAsync() |> Async.AwaitTask
  9. }
Note how we use an underscore to show that the variable value is ignored.