SerializationTests::Actions started failing recently

[j4james]

Windows Terminal version

Commit 848314e

Windows build number

10.0.19044.1706

Other Software

No response

Steps to reproduce

1. Build a recent commit of OpenConsole (I think anything after 799b5d4 should do).
2. Run the unit tests.

Expected Behavior

All the tests should pass.

Actual Behavior

There is a failure in SettingsModelLocalTests::SerializationTests::Actions when attempting to serialize the pair of findMatch actions. It seems that the commands are being reordered.

Click to see error

Error: Verify: AreEqual(toString(json), toString(result)) - Values ([
        {
                "command" :
                {
                        "action" : "findMatch",
                        "direction" : "next"
                },
                "keys" : "ctrl+shift+s"
        },
        {
                "command" :
                {
                        "action" : "findMatch",
                        "direction" : "prev"
                },
                "keys" : "ctrl+shift+r"
        }
], [
        {
                "command" :
                {
                        "action" : "findMatch",
                        "direction" : "prev"
                },
                "keys" : "ctrl+shift+r"
        },
        {
                "command" :
                {
                        "action" : "findMatch",
                        "direction" : "next"
                },
                "keys" : "ctrl+shift+s"
        }
])

I'm not sure if this is actually an indication of a bug - is the order of the commands expected to be retained? But at the very least I'd expect the test to be corrected if this is just an altered hash map order or something of that sort.

If it helps, I think the first time the test stopped working was after commit 799b5d4.

[zadjii-msft]

If it helps, I think the first time the test stopped working was after commit 799b5d4.

That's w e i r d, Cause, that shouldn't really affect this at all 🙃

I don't think the order really matters at all tbh, string comparing the json was just a shortcut vs having to actually re-parse and structurally analyze for equality. Let's just swap the order around (if it's seemingly stable in the new order)

[zadjii-msft]

I've got a fix in, well... 5a40e80 I think. I think I accidentally did a git add --all there, so it's a bit tangled with other prototypes. I'll submit the PR tomorrow.

[zadjii-msft]

Note to self: dev/migrie/b/12788-did-it-work. That's the branch where I tried to do the CI changes that really did not work.