Copying and hashing files

Question

I wonder what is more battery friendly and ultimativly more performant:

1. Copy a file and hash it then

return FileIO.readBufferAsync(sourceFile).then(function(buffer) {
            var hash, hashData;
            hashData = hashAlgo.hashData(buffer);
            hash = CryptographicBuffer.encodeToHexString(hashData);
            return sourceFile.copyAsync(folder, "" + hash + sourceFile.fileType + "-tagging", NameCollisionOption.replaceExisting).then(function(blobFile) {

I consider this variant less performant than the following, because the file is essentially read twice. Or are there any optimizations in place that a copyAsync is performing differently than a readBufferAsync + writeBufferAsync?

Also I have the problem, that  FileIO.readBufferAsync fails to load very large files (tried with 3.3 GB iso image). Is there a reason for that?

2. Read a file, build the hash from the buffer and write the buffer to disk

Answer 1

I wanted to hash the 3.3GB Windows 8 ISO image using FileIO.readBufferAsync(file).then(function(buffer).

After some time the promise fails with "Not enough storage space". I assume it means "not enough memory" but I checked in the Tastmanager and the app is not consuming any more memory after the operation has started. 

Since I need to be able to hash very large files whats the recommended way to do that?

Answer 2

Phil,

That seems to be a hard limit.  I don't think there is a workaround but have asked the product team.  I will update you on what I find out!

-Jeff

---

Jeff Sanders (MSFT)

Answer 3

Hi Phil,

The readBufferAsync method creates a copy in memory which obviously is a bad thing for such a large file.  Instead, see the documentation on streams, http://msdn.microsoft.com/en-us/library/windows/apps/windows.storage.streams.iinputstream.readasync.aspx, which allows the caller to incrementally read the file.

You have to work with the data by reading incrementally into your own buffer

-Jeff

---

Jeff Sanders (MSFT)