The C API documentation should also be updated to reflect this change.

I would be in favor of a new error code if you want to add one. Either that, or I guess we could try to make it consistent how we report the failure for these generic errors. I've found it's sometimes been difficult trying to help someone else diagnose errors like client_invalid_operation if they aren't already fairly experienced with FDB, and that probably ends up giving them a bad impression too.

I'll update the documentation and add a new error code

The C API documentation should also be updated to reflect this change.

I'm not sure how exactly to change it. It's accurate as is, and I don't want to encourage anyone to rely on the new "throws a particular error" behavior

From the C API documentation:

fdb_future_block_until_ready() will return an error only in exceptional conditions (e.g. out of memory or other operating system resources).

This is no longer accurate, you can also throw an error if you call this from the main thread.

Never call this function from a callback passed to fdb_future_set_callback(). This may block the thread on which fdb_run_network() was invoked, resulting in a deadlock.

The advice still holds, but this won't cause deadlock anymore.

I'm going to change this so that it only throws if the future is not ready, since it's possible that there are users who rely on the current behavior of fdb_future_block_until_ready not blowing up if you call it from the network thread and it's ready (even though the documentation advises against that)