WIP: S3: Implementation for S3 Get Bucket Location

[theanalyst]

S3 API supports getting the location for a bucket, which gives out one
of those geographic zones (US-WEST-1, EU for eg). Also it returns an
empty string for the default region.
(http://docs.aws.amazon.com/AmazonS3/latest/API/RESTBucketGETlocation.html)

Since LocationConstraint corresponds to regions in our case, this API
returns the region, for the "default" region empty string is returned

Signed-off-by: Abhishek Lekshmanan abhishek.lekshmanan@ril.com

[theanalyst]

@yehudasa thanks for the patient reviews so far! :) I have pushed the changes for XMLFormatter and JSON Formatter with a dump_format_ns.
(not sure of this) also I had to extend rgw_formats with the same function.. as of now I raised a NotImplementedError... should the implementation be done for this?

Also for get bucket location I have incorporated your comments changes. Let me know your comments

[yehudasa]

@theanalyst Basically ok. There are a few issues though: json doesn't define a namespace (as far as I could fine at least), so the dump_format_ns() should just do what dump_format() does. The same goes for the plain formatter (which I might have missed, but didn't see its changes here).
The base Formatter::dump_format() should just call Formatter::dump_format_ns() with an empty string. Its specializations should be removed.
Also, you can probably squash all the formatter changes together.

[theanalyst]

There are a few issues though: json doesn't define a namespace (as far as I could fine at least), so the dump_format_ns() should just do what dump_format() does.

Yeah I couldn't find anything abt json ns either.. just borrowed what json open section in ns does... I'll do what dump_format() does

The same goes for the plain formatter (which I might have missed, but didn't see its changes here).

The base Formatter::dump_format() should just call Formatter::dump_format_ns() with an empty string. Its specializations should be removed.

dump_format_ns with an empty ns should behave same as dump_format however, since this was a variadic fn and I cant pass on the varargs, I could write a helper function of sorts like which takes the namespace argument and either the buffer itself or a va_list and can be called by both? (there might be a short straight way to do this, do correct me if this is the case)

Also, you can probably squash all the formatter changes together.

Sure, will definitely do that.

[theanalyst]

Hi @yehudasa I made dump_format_ns in json & plain formatter exactly same as dump_format. Squashed the formatter changes for json & XML into one. Also added a basic unit test for XML part of dump_format_ns. Let me know your comments

[yehudasa]

@theanalyst just squash db3554a and 94b8a19 (put them after the formatter commits).

[theanalyst]

@yehudasa have squashed the commits onto d8e672f & rebased to move after formatter.