fixed updating namespaced attributes is broken

[AndyOGo]

fixes #101

Using SVG sprites with <use xlink:href="" /> will not update the icon.

The reason is that getAttributeNS and setAttributeNS DOM APIs are inconsistent in it's arguments.
E.g.: lets say 'foo:bar' ('foo' being the namespace and 'bar' the local attribute name):

• getAttributeNS only needs the localName, i.e. getAttributeNS('https://www.w3.org/1999/foo', 'bar');

  getAttributeNS(namespace, localName)
  https://www.w3.org/TR/DOM-Level-2-Core/glossary.html#dt-localname
  A local name is the local part of a qualified name. This is called the local part in Namespaces in XML [Namespaces].

• BUT setAttributeNS expects it's fully qualified name, i.e. setAttributeNS('https://www.w3.org/1999/foo', 'foo:bar');

  setAttributeNS(namespace, qualifiedName, value)
  https://www.w3.org/TR/DOM-Level-2-Core/glossary.html#dt-qualifiedname
  A qualified name is the name of an element or attribute defined as the concatenation of a local name (as defined in this specification), optionally preceded by a namespace prefix and colon character. See Qualified Names in Namespaces in XML [Namespaces].

[AndyOGo]

All tests pass:

1..567

tests 567

pass 567

ok

[AndyOGo]

@yoshuawuyts
@kristoferjoseph
@bcomnes
May I asked you to look into this as soon as you have some free time?
I'm happy to get your feedback:)

[AutoSponge]

This same code was submitted to morphdom but is a noop. All it does is rename a variable. I suggest close.

[AndyOGo]

That previous comment is incorrect.

@AutoSponge does not understand the flow of program execution during runtime and he is trolling my reputation.

[LucaMele]

@AutoSponge how can this only be a variable name change? The magic is on line 24 and 29 as attrName contains the original value instead of the conditional one removed at line 19.

[yoshuawuyts]

I'm not having any objections to this patch. Interested in hearing more from @AutoSponge though, I value his experience from prior work on morphdom.

@AndyOGo I wouldn't stress about reputation here; this patch is pretty great. If we can keep this on-topic (without any vigorous up/downvoting) I'm sure we can figure out the right approach for this patch. I'm convinced everyone is here with the best intentions.

Also would it perhaps be possible to add a test to catch any future regressions for this? Only if it's not too much effort. Thanks!

edit: also thanks for your first contribution; it's very thorough - that's real nice 😁

[AndyOGo]

@yoshuawuyts
Thanks for your quick and very friendly answer.

Btw the main contributor of morphdom is @patrick-steele-idem

[goto-bus-stop]

Ah, I see what its doing now. It looked like a noop to me first too but the difference is that attrName is unaltered (still the full path) in the set branch with this patch. @AndyOGo if you could add a test we can merge this. Because the API behaviour is so unintuitive it would be easy to accidentally break this in the future without a test