You've discovered a bug in the framework, as far as I can tell. It's reasonably subtle, because of the interaction of a few things:
- When you call
ReadLines(), the file is actually opened. Personally, I think of this as a bug in itself; I'd expect and hope that it would be lazy - only opening the file when you try to start iterating over it.
- When you call
GetEnumerator() the first time on the return value of ReadLines, it will actually return the same reference.
- When
First() calls GetEnumerator(), it will create a clone. This will share the same StreamReader as textEnumerator
- When
First() disposes its clone, it will dispose of the StreamReader, and set its variable to null. This doesn't affect the variable within the original, which now refers to a disposed StreamReader
- When
Last() calls GetEnumerator(), it will create a clone of the original object, complete with disposes StreamReader. It then tries to read from that reader, and throws an exception.
Now compare this with your second version:
- When
First() calls GetEnumerator(), the original reference is returned, complete with open reader.
- When
First() then calls Dispose(), the reader will be disposed and the variable set to null
- When
Last() calls GetEnumerator(), a clone will be created - but because the value it's cloning has a null reference, a new StreamReader is created, so it's able to read the file with no problems. It then disposes of the clone, which closes the reader
- When
GetEnumerator() is called, a second clone of the original object, opening yet another StreamReader - again, no problems there.
So basically, the problem in the first snippet is that you're calling GetEnumerator() a second time (in First()) without having disposed of the first object.
Here's another example of the same problem:
using System;
using System.IO;
using System.Linq;
class Test
{
static void Main()
{
var lines = File.ReadLines("test.txt");
var query = from x in lines
from y in lines
select x + "/" + y;
foreach (var line in query)
{
Console.WriteLine(line);
}
}
}
You could fix this by calling File.ReadLines twice - or by using a genuinely lazy implementation of ReadLines, like this:
using System.IO;
using System.Linq;
class Test
{
static void Main()
{
var lines = ReadLines("test.txt");
var query = from x in lines
from y in lines
select x + "/" + y;
foreach (var line in query)
{
Console.WriteLine(line);
}
}
static IEnumerable<string> ReadLines(string file)
{
using (var reader = File.OpenText(file))
{
string line;
while ((line = reader.ReadLine()) != null)
{
yield return line;
}
}
}
}
In the latter code, a new StreamReader is opened each time GetEnumerator() is called - so the result is each pair of lines in test.txt.