assertXMLEqual fails when white space is present outside XML fragment

[Sergiy Kuzmenko]

Currently this test fails:

class MyTestCase(TestCase):
    def test_xml(self):
        self.assertXMLEqual('<greeting>hello</greeting>', '<greeting>hello</greeting>\n')

I believe this is a bug because white space outside root tag should not matter. It makes testing rather confusing when reading XML documents from files (which may or may not have trailing new line characters).

[Tim Graham]

Seems reasonable as the docstring says "Whitespace in most cases is ignored".

[Mihai Iachimovschi]

At the time of analyzing the issue, I didn't notice that the issue is already assigned. Anyway, just because the fix seems so simple, and there is no pull request yet, I published my own solution as a PR.

The fixed branch is available here: ​https://github.com/mishunika/django/tree/ticket_25170
And the PR is here: ​https://github.com/django/django/pull/5049

[Kamil Warguła]

I started working on this bug after assigned.
This is PR for this issue: ​https://github.com/django/django/pull/5052
Resolution of this bug is different then @mishunika.

[Tim Graham]

I think ​PR 5052 is more correct in that it will continue to consider whitespace *inside* a tag significant; e.g. the following snippets shouldn't be considered equal:

<tag>foo </tag>
<tag>foo</tag>

Anyone to confirm that? The author of ​PR 5049 argues otherwise.

[Jacek Bzdak]

Whitespace around tags is indeed siginificant --- technically following documents *are* different

<foo></foo>

and

<foo> 
</foo>

however for most practical purposes this difference is insignificant, as most of XML-based formats ignore whitespace altogether (like html). Despite that I think that in unittesting framework we should rather assume pessimistic case that whitespace does matter.

[Mattia Larentis]

According to ​https://tools.ietf.org/html/rfc3470#section-4.16

In XML instances all white space is considered significant and is by default visible to processing applications

So, I think that @jbzdak is right.

[Mattia Larentis]

The fixed branch is available here: ​https://github.com/nostalgiaz/django/tree/pull_25170
And the PR is here: ​https://github.com/django/django/pull/5580

[Simon Charette]

Contained white-space should really be considered significant in the HTML case.

If I remember correctly it affect display: inline elements in some way (or it did in the past).

Stripping leading and trailing white space makes sense to me.

[Tim Graham <timograham@…>]

In 2085d8d5:

Fixed #25170 -- Made assertXMLEqual()/assertXMLNotEqual() ignore leading and trailing whitespace.

Thanks Jacek Bzdak for indepdently contributing a similar fix.