Possible glitch in the code

Oct 18, 2012 at 11:36 AM


First thanks for this amazing piece of code it's really good and clear. As I'm not a .net user I decided to get inspiration from it to do the same kind on Java. Is that ok?
While reading your code I've noticed a little glitch in the code (it won't cause bug). On the AzureIndexInput on the Close() method you do:

blobStream = new StreamInput(CacheDirectory.OpenInput(fileName));

Wouldn't it be much more coherent to just do:

blobStream = _azureDirectory.OpenCachedInputAsStream(fileName);

As it does the same thing (I think so) and what is also used in AzureIndexOutput.

This is not much but I would also like to be sure this is right.

Finally I don't think you will be able to help me as you might not be familiar with Java Lucene (even though they seem really close). But I'm having issues with my .nrm files that always make my RAMInputStream read past EOF. They are always 5 bytes. The weirdest thing is this happens when in the call blob.upload() so I don't think it actually comes from Lucene. Anyhow did you ever get any issues with .nrm files?  


Thanks again for your really good code!