.Count() is dangerous

[defcon84]

Not really an IPAddressRange issue, but maybe an IEnumerable issue.

But if you use IPAddressRange.Count() on a large list of IPs, it will perform bad and take minutes.

for example: IPAddressRange.Parse("0.0.0.0/1").Count()

or even worse: IPAddressRange.Parse("0::0/0").Count()

[defcon84]

maybe something like:

public double Count()
{
    byte[] byteBegin = this.Begin.GetAddressBytes();
    byte[] byteEnd = this.End.GetAddressBytes();

    double sum = 1;
    for (int i = byteBegin.Length - 1;i >= 0;i--) {
        if (byteEnd[i] - byteBegin[i] == 0)
            continue;

        sum += (byteEnd[i] - byteBegin[i]) * Math.Pow(2D, (double)((byteBegin.Length - 1) - i) * 8);
    }

    return sum;
}

[jsakamoto]

Hi, @defcon84,
Thank you for your issue.
But I would like to do not any actions about this issue. This is by design.

Dangerous LINQ operations are not only Count() method. ToArray(), ToList(), and also simple foreach enumeration are dangerous too.
Many lazy evaluation systems have this problem, not only the IEnumarable interface of this library.

Of course, we can avoid this issue by implement custom Count() method such as your example code.

But, I feel it is over spec about implement custom Count() method on this library.

This is my answer.
Would you tell me your opinion, @defcon84 or anybody else?

[defcon84]

Yes, as i said in my first message, i know this is not the issue of this library. And maybe it's best not to implement it. I just wanted to warn people and give them an alternative for it.

Thank you for this library!

[jsakamoto]

Hi @defcon84 , thank you for your reply.

I just wanted to warn people and give them an alternative for it.

Oh, sorry, it's my mistake of reading your post. Thank you for your useful post :smile:

[defcon84]

no problem :)

and thank you!