RUST-2097 Migrate from derivative to derive-where
#1245
Conversation
| crypt: Crypt, | ||
| exec: CryptExecutor, | ||
| internal_client: Option<Client>, | ||
| _internal_client: Option<Client>, |
There was a problem hiding this comment.
I got a dead code lint error for this field when I switched the Debug impl on the struct, but deleting it caused errors 🤔 I think something else is holding a weak reference to it?
There was a problem hiding this comment.
The CryptExecutor holds a WeakClient, so that's probably it. I vaguely recall that it's weak because whether it's using internal_client or a user-supplied Client depends on whether the context is implicit or explicit encryption, and if it's user-supplied we don't want to hold a strong reference and keep it alive longer than the user expects.
| crypt: Crypt, | ||
| exec: CryptExecutor, | ||
| internal_client: Option<Client>, | ||
| _internal_client: Option<Client>, |
There was a problem hiding this comment.
The CryptExecutor holds a WeakClient, so that's probably it. I vaguely recall that it's weak because whether it's using internal_client or a user-supplied Client depends on whether the context is implicit or explicit encryption, and if it's user-supplied we don't want to hold a strong reference and keep it alive longer than the user expects.
Can you file a ticket so we remember to do this? |
isabelatkinson
changed the title
derivative to derive-wherederivative to derive-where
|
Can we make a minor release of this |
|
@TroyKomodo 3.1.1 was just released! |
Migrates from
derivativetoderive-where. In the long term, I'd like to switch over toderive_more, which we already depend on for someDisplay-related utilities.derive_morehas the largest feature set of crates of its ilk; however, its MSRV for the features we'd need is 1.75 (compared to our current 1.67), and it does not yet supportPartialEq. We can make the swap once those concerns are resolved. This is tracked in RUST-2102.