[MRG+1] support namespace prefix in xmliter_lxml

[tpeng]

No description provided.

[pablohoffman]

why not add the prefix as the last argument (to maintain backwards compatibility with positional arguments)?

[tpeng]

because it's called prefix :), so i thought it's better put it before the namespace

[pablohoffman]

It doesn't sound like a strong enough reason to break backwards compatibility, I'm happy to merge if you leave prefix argument after namespace.

[kmike]

@tpeng I wonder why are you using this code. We have some evil plans to remove it from Scrapy (see #1063), or to move it to some other place.

But I'm not familiar enough with xmliter_lxml - is it really better than scrapy.utils.iterators.xmliter? If so, does it make sense to move it to scrapy.utils.iterators? Or maybe even use it by default?

[pablohoffman]

@tpeng bump to @kmike question

[tpeng]

@kmike sorry for the delay. the original reason was i found the xmliter_lxml is memory efficient and can work with the xml i used. scrapy.utils.iterators.xmliter didn't work if the namespace is not empty.

[kmike]

See also: #605 which introduces a third way to parse XML iteratively in Scrapy.

[nyov]

In #1063 it was decided to keep xmliter_lxml in favor of xmliter.
Let's merge this before the relocation.
(I'll update #1134 afterwards)

[kmike]

+1 to merge it.