code review 66850043: encoding/xml: Marshal nested namespaces correctly.

encoding/xml: Marshal nested namespaces correctly.

Fixes issue 7113.

[anacrolix, 2014-02-25 03:08:25]

Hello dave@cheney.net (cc: golang-codereviews@googlegroups.com),

I'd like you to review this change to
https://code.google.com/p/go

[anacrolix, 2014-03-03 04:41:25]

Hello golang-codereviews@googlegroups.com,

Please take another look.

[gobot, 2014-05-09 21:25:35]

R=rsc@golang.org (assigned by iant@golang.org)

[rsc, 2014-05-13 04:32:40]

Until now Encoder has been an opaque type. I would like to keep it that way. In
any case this CL contains no documentation about how to use the default ns, and
it seems to be up to the user of Encoder to keep it correct. None of this seems
right to me. The encoder gets the name spaces on attributes correct, including
prefixes, automatically. It should do the same here. 

CL 93320043 is my counter-proposal, but I am worried about backwards
incompatibilities in the encoding. I'd like to wait until 1.4 to submit it.

[anacrolix, 2014-05-13 22:55:29]

On 2014/05/13 04:32:40, rsc wrote:
> Until now Encoder has been an opaque type. I would like to keep it that way.
In
> any case this CL contains no documentation about how to use the default ns,
and
> it seems to be up to the user of Encoder to keep it correct. None of this
seems
> right to me. The encoder gets the name spaces on attributes correct, including
> prefixes, automatically. It should do the same here.

I think my concern was that sometimes it's necessary to generate an XML snippet
to be embedded in existing XML, where the default NS is not "". In these cases I
think without being able to specify the default NS for a call to Marshal,
redundant xmlns attributes will be generated.

> CL 93320043 is my counter-proposal, but I am worried about backwards
> incompatibilities in the encoding. I'd like to wait until 1.4 to submit it.

Your CL looks much better, thanks.

[dfc, 2014-07-16 10:57:55]

R=close

Replaced by 93320043