Why is traversal forward only most of the time?

I have noticed that many iterators or data readers are forward only like DataReader, XmlReader, IEnumerator, any many more (you got the idea).

So simply asking why are they forward-only usually when i create a Data Iterator for my custom needs i usually Try adding support for navigation on both side开发者_如何学运维s. I agree that most of time we don't need backward traversing but sometimes we do need and so we end up creating temp variables or something to hold on the data while required.

So my questions are:

• Why are most Data-Iterators forward-only

• Am i wrong in creating a backward traversable Iterator/ Data reader. if not why doesn't framework have such support for its inbuilt Data Iterators.

• Do we have any serious performance drawback or its just not considered a good design to have such feature.

This question has bugged me a lot from start but never got a satisfactory answer so I'm asking it here.I do believe many developers may agree with me that backward traversing can be useful sometimes.

"forwards only" is:

• the most common use for most consumers
• simple to implement, so the most likely to be implemented by most producers
• the only thing we can guaranteed if we assume we don't want to buffer all the data in memory
• easily buffered to allow random access (for moderate sized data)

for example, if you are reading data from a database, a network stream, etc, you can only guarantee "forwards". We certainly don't want to arbitrarily buffer all that data - it could be huge potentially.

If the client thinks they have a sane amount of data, they can always call ToList() etc to buffer it in memory and allow random access.

For example, consider this perfectly valid sequence:

public static IEnumerable<int> LotsOfData() {
    var random = new Random();
    while(true) yield return random.Next();
}

• it can't be reversed without buffering
• it is infinite in length, so can't be buffered

obviously that example is a little unlikely, but reading from a socket, database, or even a large file - can be essentially the same scenario.