encoding/xml: no way to extract value of tag when namespaced tag with same name occurs later

[evmar]

Briefly, there doesn't appear to be a way to get "bar" out of the following
XML:

<x>
<foo>bar</foo>
<unrelated:foo>...</unrelated:foo>
</x>

Matching on `xml:"foo"` grabs the latter unrelated tag.

Here's a runnable example exhibiting the problem:
  http://play.golang.org/p/m-17dw-fDn

This problem affects gdata feeds (Google APIs) which make heavy use of XML namespaces.

[rsc]

Comment 1:

Labels changed: added priority-later, go1.1, removed priority-triage.

Status changed to Accepted.

[rsc]

Comment 2:

Labels changed: added go1.1maybe.

[rsc]

Comment 3:

Labels changed: removed go1.1maybe.

[rsc]

Comment 4:

Labels changed: added size-l.

[tianon]

Comment 5:

Having this same issue trying to parse RSS from Slashdot.  They include <link> as
per the standard, then later include <atom10:link>, and no global xmlns on the
main <rss> tag to help disambiguate the two, so the first literally has no
namespace at all.
I tried several things to get around this, including using `xml:" link"` to explicitly
specify to match a blank namespace (which was a long shot, obviously), and then found
that the code I'm looking for is src/pkg/encoding/xml/read.go on line 242 which
(rightfully) ignores an empty namespace so that if we don't care about namespaces, we
don't have to specify them.
I'm not exactly sure what a solid fix for this might be, but figured I would add an
extra couple cents here.
My only idea is to add a special case namespace string that intentionally matches the
case of an empty namespace (thus a tag like `xml:"_ link"`), but that might conflict
with using something like xmlns:_="http://example.com" or xmlns:something="_" (which I
freely admit that I don't know if are valid, thus whether really are problems).

[tianon]

Comment 6:

Did a little digging and found out that technically speaking,
xmlns:_="http://example.com" is valid. (see
http://www.w3.org/TR/REC-xml/#NT-NameStartChar)
Not sure what simple alternative could be provided, but figured I'd add a little
followup to my earlier idea.  I also am not sure whether or not that specific example
has enough real world use to have an appreciable negative impact.
This cursory Google search (https://google.com/search?q="xmlns%3A_") suggests that at
least one person is using it, but there really aren't all that many hits, for what it's
worth.

[tianon]

Comment 7:

Also for what it's worth, a namespace name can't start with tilde or period, so those
could be acceptable alternatives (especially since hyphen is already taken for ignoring
a field).

[gopherbot]

Comment 8 by cornelius.howl:

Having this issue when parsing appcast.xml file (which has sparkle:version ..
attributes) can not be parsed.......

[rsc]

Comment 9:

https://golang.org/cl/7227056/

[rsc]

Comment 10:

This issue was closed by revision 4dd3e1e.

Status changed to Fixed.