Use XML::Hash::XS

[slobo]

This makes t/10_responses.t pass

[slobo]

So, in this first iteration, we learn that SuppressEmpty => undef behaviour is the only difference that test suite can uncover for 10_responses.t. But, while doing this in perl code is probably fast enough for smaller results at least, it feels rather dirty, so I'll try to petition XML::Hash::XS to add equivalent behaviour.

[slobo]

@pplu, with this change, test suite passes on my machine (minus some failure on 01_load.t, but same happens on the base branch, and I hope is unrelated). Let me know if there is anything else we need to do. Thanks!

[castaway]

Looks like we should look at merging this for 0.44 (before 0.45), see #372

[byterock]

Hmm might clean up the issue I had with a few of the S3 actions where the parsing the XML beheaded it and I had to add in a little extra logic to handle the missing root key.

Would have to look at my notes to see which one.

[castaway]

There are a few responses that don't contain the main XML key (there's code in RestXMLResponse to handle it) - would be handy to get rid of XML::Simple regardless

[slobo]

@pplu , this PR just merges into your xml2hash branch, so probably safe to accept and continue work in there?

[slobo]

Or, should we maybe change the target to be the next release branch to see how it behaves?

[castaway]

working into xml2hash seems sane to me, I hadnt seen it was against that one

[mschout]

Would love to see this stop-gap make it into a release.