1.5.2 breaks serialization for formats like "bincode"

[nullchinchilla]

Serializing a Hash produces a bytestring instead of an array in
formats that support bytestrings (like CBOR). Deserialization is
backwards-compatible with the array format.

This is not backwards compatible for formats like "bincode", where it's not possible to first try to deserialize as an array and then try deserialization as a bytestring. Previously serialized and saved Hashes will now fail on deserialization.

This has actually broken a large production system I run that currently must be pegged to 1.5.1 as a result. Please at least make this feature optional, or enable it only for human-readable serde formats like JSON.

[oconnor663]

I'll revert this change tonight, and then we can consider whether there's a better way to accomplish the same thing. Do you have time to put together a small program that reproduces the break you're seeing?

[oconnor663]

I've published v1.5.3 with the revert. Let me know if you can confirm that it fixes your system.

[nullchinchilla]

Yes it does fix it!

I think a better way of doing this is probably to check whether the format is human readable, which might be a good proxy for whether it's self-describing.

But I think probably the best way is to release 2.0 instead.

[BurningEnlightenment]

I think a better way of doing this is probably to check whether the format is human readable, which might be a good proxy for whether it's self-describing.

CBOR, JSONB and MessagePack would be missed by that check, right?