GrizzlyResponse.getResponseBodyAsBytes() implemented very inefficiently about async-http-client

[cowtowncoder]

(note: hope to provide patch for this in a day or two)

[rlubke]

I'm spending today working on my AHC issues.

Thanks for spotting this. It was a quick and dirty impl to get things going.

A better implementation would be:

final byte[] responseBodyBytes = new byte[responseBody.remaining()];
final int origPos = responseBody.position();
responseBody.get(responseBodyBytes);
responseBody.position(origPos);
return responseBodyBytes;

If you're fine with this as well, then I'll commit, otherwise let me know what you're thinking.

[cowtowncoder]

Hmmh. I am also working on patches, I hope I can merge things... there is LOTS of clutter, wrt replication on different impls. And grizzly is actually cleanest of them (this was about the only major thing I noticed).

How about I try merging in my changes, and you can add this patch (which is better than what I had)

[cowtowncoder]

Merged in my changes -- I'd appreciate if you have a look at what I did, since I didn't feel fully comfortable with most optimal changes wrt Buffer handling. So you can probably further improve code; and that'd be great overall.

[rlubke]

Will review the changes on the master. I've implemented the change above in the 1.7.x branch.

[rlubke]

I've made some small modifications on master and ported the getResponseBodyAsBytes().

Will close this issue out.