使用组合属性时，Visual Studio下的NUnit极其慢

Question

My set up is NUnit (latest, 2.6.something), VS13 with R#8 and Win7. I noticed to my great surprise that when I'm using the attribute Combinatorial , the IDE gets extremely, painfully slow. We're talking waiting at the greyed out window of VS for a few seconds after each keystroke !

I've googled that but there seems not to be any relevant hits. I suspect our local setup to blame (how one can fail to install NUnit and R#?!) but it could be of interest for others to know what to do when that happens.

We've noticed that the delay increases exponentially as we increase the number of test cases. This is an example for one of the methods.

[Test]
[Combinatorial]
public void AddWaveformTest(
  [Values(null, "", "sec", "stringThatMayBulky")] string s1,
  [Values(null, "", "sec", "stringThatMayBulky")] string s1,
  [Values(null, "", "sec", "stringThatMayBulky")] string s3,
  [Values(null, "", "sec", "stringThatMayBulky")] string s4,
  [Values(int.MinValue, int.MaxValue, -1, 0, 1)] int i1,
  [Values(null, "", "sec", "stringThatMayBulky")] string s5,
  [Values(null, "", "sec", "stringThatMayBulky")] string s6,
  [Values(null, "", "sec", "stringThatMayBulky")] string s7,
  [Values(null, "", "sec", "stringThatMayBulky")] string s8,
  [Values(int.MinValue, int.MaxValue, -1, 0, 1)] int i2)
{
  ...
}

That's a bit above 26 millions operations, which might be a lot but in computer terms it shouldn't be that bad. Right? I also noticed that it's not that much the Combinatorial attribute that is the direct cause but Values attribute. When I comment out the latter, stuff work again, while commenting out the former brings no improvement...

Answer 1

You may want to check this thread as it looks like resharper may have a bug in a 8.x version that causes the problem of the editing window taking some time after each keystroke to respond.

Some areas that were put aside in the comments to the question are:

• setup and teardown being expensive: this idea was tied with the second point ->
• some sort of continuous testing exists inside Resharper (sthing like ncrunch capabilities, or approching) but OP doesn't find it likely.

Don't hesitate to come and comment on the problem, I'm very curious about it (btw, does the time during which the editor window is unavailable matches your test duration approximately?)