Commandfor buttons with missing or invalid type attributes currently work as command buttons

[lukewarlow]

What is the issue with the HTML Standard?

Per #10832 the intention was that implicit submit buttons would be a no-op but the spec currently makes them work as command invokers.

cc @keithamus

The spec also says that buttons with an invalid type attribute works as a command button. Previously I believe the intention was that these would continue to be submit buttons (though I may be misremembering that).

We should correct the spec, and ensure the tests match for both the missing and invalid cases.

[zcorpan]

Hmmm, I would prefer to make the UA behavior be what we want to happen long term, but use document conformance rules to encourage adding type="button". Otherwise authors will have to include type="button" for far longer than 5 years.

[lukewarlow]

Cc @keithamus and @annevk

We can perhaps discuss this at whatnot tomorrow? I don't mind either way we can update implementations and the tests relatively easily.

[annevk]

Only when the state is Auto and there is a form owner we would require type=button for now. That still seems reasonable to require and move away from that in a couple of years once it's widely implemented.

[lukewarlow]

#11053 - I've got a PR based on discussions, it removes reset buttons (inc inputs) working as popover or command buttons, it also stops auto buttons acting as command buttons they now no-op as intended originally (I would like to get the spec in a better state and then we can separately discuss if they should no-op or not in a follow up PR)