Create a simple example for the IgnoreFields

[colinnewell]

I was struggling to figure out how to use IgnoreFields initially so I figured an example would be helpful. I've essentially ripped off the example from the parent module with a proper copy paste job for now with a quick example.

If you can suggest a better way to do it, I'd be happy to clean up the code.

I suspect I could also make the example clearer, but I'd quite like to clean up the code before going too far.

[frioux]

This seems like a solid PR to me. The only problem is that cmp explicitly thwarts tests like this by injecting randomness into the output format. @colinnewell you might set the flag that disables that for your test and then it'd pass.

[colinnewell]

Thanks for the comments @frioux. Did you see where this randomness is coming from? I'm failing to spot it so far. One of the reasons I didn't push this PR too far was that it was pretty much a rip off of cmp/example_test.go which seemed ugly. The other thing that I'm struggling to figure out is why that one doesn't suffer from the same problem mine does.

If figure I may as well try to get the tests to pass consistently rather than randomly.

[colinnewell]

Ah, found it: cmp/report_text.go.

[colinnewell]

Thanks for the prod. I've rebased and fixed the test so that it runs consistently.

[dsnet]

Thanks!

[dsnet]

will still potentially show ignored fields as different,

Is still true? I recall fixing the reporter to stop printing ignored fields and using ... instead, even for adjacent fields.

[dsnet]

go-cmp/cmp/report_compare.go

Lines 258 to 265 in d713870

	// Elide ignored nodes.
	if r.Value.NumIgnored > 0 && r.Value.NumSame+r.Value.NumDiff == 0 {
	deferredEllipsis = !(k == reflect.Slice || k == reflect.Array)
	if !deferredEllipsis {
	list.AppendEllipsis(diffStats{})
	}
	continue
	}

[colinnewell]

The example itself demonstrates what I was trying to say. I ignore IPAddress but it still shows up in the diff. I assume that the reason I added 'potentially' to the sentence is because I had come across your behaviour to minimise that.

If you do end up changing that behaviour, the example will highlight it so it should be easy to change that comment then.

Alternatively, is it the word 'adjacent' that you're objecting to?

[dsnet]

I think there's a crucial difference between "in the diff" as you just said versus "as different" as currently written in the example. The former denotes an issues with display representation, while the latter subtly suggests something about the semantics of how the comparison is performed.

Perhaps this is better worded as:

// While the specified fields will be semantically ignored for the comparison,
// the fields may be printed in the diff when displaying entire values
// that are already determined to be different.

[dsnet]

Please add license header.

[colinnewell]

Done

[dsnet]

I know people aren't fond of 2020, but that's the current year ;)

[dsnet]

I know people aren't fond of 2020, but that's the current year ;)

[dsnet]

I think there's a crucial difference between "in the diff" as you just said versus "as different" as currently written in the example. The former denotes an issues with display representation, while the latter subtly suggests something about the semantics of how the comparison is performed.

Perhaps this is better worded as:

// While the specified fields will be semantically ignored for the comparison,
// the fields may be printed in the diff when displaying entire values
// that are already determined to be different.

[dsnet]

// Use IgnoreFields to ignore fields on a struct type when comparing
// by providing a value of the type and the field names to ignore.
// Typically, a zero value of the type is used (e.g., foo.MyStruct{}).

[colinnewell]

Used, thanks.