Cannot reproduce Apollo Client 2 cache-and-network fetch policy about apollo-client HOT 3 OPEN

Hi, which apollo client version are you using, currently?

from apollo-client.

Hi Olivier,

you are right that our recommendation to reproduce this old behaviour is to start with cache-and-network and then use a nextFetchPolicy of cache-first.

Your query will immediately move over to the new fetchPolicy once the request is started, so that gets me thinking that this might not be your problem here.

There are situations where even with a fetchPolicy of cache-first, a mutation triggers an outgoing query. Namely, if the mutation (or another query) overwrites a part of the cache with data that is not sufficiently complete to answer your initial query.

That would usually happen if your request and mutation somehow overlap, and for some reason your cache cannot be normalized.

When you inspect the cache in the devtools, is is normalized properly? If not, you might be missing id fields in your queries in some places 🤔

from apollo-client.

Hi,

Ok that clears things up a bit. Makes me think that the change in behaviour is then likely caused by the change in InMemoryCache merge, but this will likely require a lot more investigation on our side because we were relying on that mechanism in a lot of places. I believe this will prevent us from upgrading to Apollo Client 3 in the foreseeable future.

Thanks!

from apollo-client.