check if byte array is list before converting to utf-8

[walterbender]

Since F19, the keyboard section of the cpsection fails when converting a byte array to utf-8. This is because the byte array isinstance list instead of isinstance string. Not sure of the root cause of this change, but this patch checks the type and does a proper conversion in either case. (See http://bugs.sugarlabs.org/ticket/4539)

[dnarvaez]

We need to understand what is going on here. The original _item_str looks weird to start with.

[dnarvaez]

Oh see http://cgit.freedesktop.org/libxklavier/commit/?id=115a22dfff5ea59cdcf8a5a7f8e15ffd71164c1c

[dnarvaez]

I suppose we should depend on 5.4 and use those accessors.

[dnarvaez]

Though Fedora 19 doesn't have 5.4. Maybe see what Peter thinks about getting 5.4 in F19? The other distros doesn't even have 5.3 which I think we need for other reasons, so it's not going to make things worse there.

If we can get 5.4 in Fedora I'd just use the accessors, otherwise I guess we need to keep a fallback doing conversions.

[dnarvaez]

I can build 5.4 in sugar-build btw, but lets first decide the approach we are taking.

[walterbender]

I think using the xklavier accessors makes sense; we can get rid of the kludgy helper method all together.

[dnarvaez]

Closing then.