Task.Run only runs synchronously?

Question

I'm trying to write a method that reads each page of a pdf, but since it takes a significant amount of time to read each page through the api and I'm looking at pdfs hundreds of pages long, I want to queue the reading of each page asynchronously and then return the results when they're ready, so multiple pages are being read at once.

I use Task.Run to "queue" the task, and I expect to see the Debug log print the pages out of order, but they only execute in order, so I think they are being run synchronously. Any ideas?

var tasks = new List<Task>();
foreach (Page page in _pdfDoc.GetPages()) {
    var task = Task.Run(() => {
        //tried adding await Task.Yield() here, doesn't work
        Debug.WriteLine("searching page " + page.Number);
        if (page.Text.Contains(query)) {
            pagesWithQuery.Add(page.Number);
        }
        howManySearched += 1;
        Dispatcher.UIThread.InvokeAsync(() => {
            searchProgress.Value = howManySearched;
        });
        return Task.CompletedTask;
    });
    tasks.Add(task);
    // await task; <== does nothing??
}
// await Task.WhenAll(tasks); <== also nothing

Answer 1

I use Task.Run to "queue" the task, and I expect to see the Debug log print the pages out of order, but they only execute in order, so I think they are being run synchronously.

You don't have enough data to support this assumption. You are only logging the starting of each Task:

Task task = Task.Run(() =>
{
    Debug.WriteLine("searching page " + page.Number);

...but you have no idea when the Task completes. You could get a better idea about the level of concurrency achieved by your code, by doing something like this:

object locker = new();
int concurrencyCounter = 0;
int maxConcurrency = 0;

Task task = Task.Run(() =>
{
    int concurrency = Interlocked.Increment(ref concurrencyCounter);
    lock (locker) maxConcurrency = Math.Max(maxConcurrency, concurrency);
    try
    {
        Debug.WriteLine("searching page " + page.Number);

        // Do work with the PDF page...

    } finally { Interlocked.Decrement(ref concurrencyCounter); }
});

//...

await Task.WhenAll(tasks);
Debug.WriteLine($"Maximum concurrency: {maxConcurrency}");

As a side note, there is a ton of idiomaticity in your code. You are not taking advantage of either the Parallel class or the AsParallel PLINQ operator, nor the Progress<T> class, and I suspect that there are also race conditions around the use of the undefined variables howManySearched and pagesWithQuery.