Michal Nazarewicz

> I'm leaning towards BorshEncoding + base64. I’m still tempted to switch from String to Vec eve though we would need to be careful not to send those messages to...

> Would this mean that ViewRuntimeAdapter::view_state would need to have a parameter that flags whether the proof should or not be included? Yes. And also QueryRequest::ViewState and possibly some other...

OK, sorry, I keep just throwing ideas some of which may make no sense, but it just ocured to me that maybe we’re thinking about it wrong? Rather than adding...

That’s this comment: --- Since we want to proof of the prefix key rather than keys we’re looking up, I think we actually want to add code for getting the...

How about this: ```diff core/primitives/src/views.rs | 11 ++++---- core/store/src/trie/iterator.rs | 32 ++++++++++++++++++++-- core/store/src/trie/mod.rs | 17 +++++++++--- core/store/src/trie/state_parts.rs | 16 +++++------ .../src/tests/runtime/state_viewer.rs | 2 +- runtime/runtime/src/state_viewer/mod.rs | 7 +++-- 6 files...

Well, yes, failures with test_view_state are expected since I didn’t update the state there so the tests expect it to be empty when it isn’t. My diff was by no...

I’m still going to be adding pytest tests but I think this can be closed.

This would give check_network_identifier a place to live. The counter argument is that not all handlers use all of those objects and having genesis field means we’ll have an Arc...

Genesis is possibly a huge structure (it includes genesis records) so we don’t want to clone it into a new structure.

Double indirection is just a minor issue. The problem is that Genesis is potentially a large object. It’s small on mainnet so we could just shrug it off but on...