Junit4: Asserts for CharSequence contents

[sabi0]

A lot of APIs in Android deal with CharSequence. And using assertEquals() with CharSequence arguments is cumbersome and unstable. One has to reproduce the exact same type which might change later.
And sometimes it just does not work (e.g. StringBuilder uses default Object.equals() implementation).

Therefore I would suggest to add new methods like
assertContentEquals(CharSequence expected, CharSequence actual)
that would compare just the content, irregardless of the type it's been wrapped into.

(Pull request #949 was trying to do that with generic assertEquals() which is indeed questionable)

I can create a pull request if you find this proposal reasonable.

[kcooney]

Important issue, but I would prefer to do it in assertEquals or an overload of assertEquals. Having to know which assert message to use for which use case is error-prone.

You might want to consider using assertion frameworks with a fluent API (like Truth or Fest).

[sabi0]

The problem with assertEquals is that sometimes you do need to compare the types too.
So I believe assertEquals("abc", new StringBuilder("abc")) must fail. Therefore the need for separate methods to compare just the content.
This is IMO very similar to the current assertArrayEquals approach.

I agree that picking the right method would become a bit trickier. But it's anyway better than the current lack of proper CharSequence support.

[kcooney]

Could you list what specific methods you are proposing?

[sabi0]

So far I've been thinking of just these two:
Assert.assertContentEquals(CharSequence expected, CharSequence actual)
Assert.assertContentEquals(String message, CharSequence expected, CharSequence actual)

[kcooney]

Could you give better examples? I would never write assertEquals("abc", new StringBuilder("abc")) (I would call build() on the builder)

Also I do think DSL based frameworks are often a better fit. In Truth I would write:

assertThat(myCharSequence).hasValue("hello");

Note that hasValue() takes a String so there is no question about what it means for two CharSequence values to be considered equal.

[sabi0]

Here is an example of how it might look in a test:
assertContentEquals("abc", bundle.getCharSequence(Notification.EXTRA_TEXT));

Adding also a HasValue Matcher is a good idea.

[kcooney]

Thanks. Since CharSequence doesn't provide a contract for equals() how do you propose we consider two CharSequence values equal? Is it simply that they have the same toString() representation? If so why not make the expected value a String?

[sabi0]

Good point. Though I am hesitant because having expected and actual values with the same type looks cleaner.

And I would also compare the lengths first. To avoid calling toString() unless it's really necessary.

[joseph-mccarthy]

I would like to take this feature request, please. Could it please be assigned to me. Thank you.

[kcooney]

@joseph-mccarthy GitHub apparently only lets you assign issues to someone on the project's team. But we almost never assign issues anyway. Feel free to submit a pull request after reading https://github.com/junit-team/junit4/blob/master/CONTRIBUTING.md

[btrajkovski]

@kcooney I have question about this issue, is it still up for grabs?
I see that it is linked in PR #1439, but that PR is in request change for quite a long time. Can I take that as a basis for a new PR and go over the review comments?

[kcooney]

@joseph-mccarthy are you still working on this?

[btrajkovski]

@kcooney I am assuming that @joseph-mccarthy is no longer working on this and I've opened a PR with implementation.