Corner cases about child tries are unexplained

[tomaka]

Some questions are in my opinion not really clear in the spec:

• What happens if the runtime calls the ext_storage_get_version_1 or ext_storage_read_version_1 host function with a key K, where K is in fact a child trie? As far as I understand, this would return the trie root hash of the child trie, but this isn't really clear in the spec.
• What happens if the runtime calls the ext_storage_set_version_1 host function with a key K, where K is in fact a child trie? I assume that it destroys the child trie, but this isn't really clear.
• What happens if the runtime calls any of the ext_default_child_storage_* host functions with a child_storage_key that is in fact not a child trie? I assume that this is the equivalent of a runtime panic?

[lamafab]

Fixed in #554

[lamafab]

Test coverage in https://github.com/w3f/polkadot-tests/pull/241, should be merged, soon

[tomaka]

I don't see how https://github.com/w3f/polkadot-spec/pull/554 covers any of the questions that I've raised, and I don't see any answer to them in the spec.

For some reason I don't have a "reopen" button, but if I had one I'd reopen the issue.

[tomaka]

I would say that the first two points are now covered by https://github.com/w3f/polkadot-spec/issues/577 and https://github.com/w3f/polkadot-spec/issues/575

Only remains:

What happens if the runtime calls any of the ext_default_child_storage_* host functions with a child_storage_key that is in fact not a child trie? I assume that this is the equivalent of a runtime panic?