I still think that we should modify this whole stack so that a result of processMessageStoreOperations could be used as a new value of messageStore. We can e.g. modify the old messageStore and pass it as an argument to processMessageStoreOperations function.

In D4201#121171, @palys-swm wrote:

I still think that we should modify this whole stack so that a result of processMessageStoreOperations could be used as a new value of messageStore. We can e.g. modify the old messageStore and pass it as an argument to processMessageStoreOperations function.

That's the goal, but getting it to that state will happen later in this stack.

If I understand correctly, I think you're suggesting that instead of ignoring the messageStore.messages from freshMessageStore(...) and substituting in the processedMessageStore.messages from the call to processMessageStoreOperations(...)... we should instead pass the output of freshMessageStore(...) into processMessageStoreOperations(...) and return the resulting messageStore.

I think that makes sense, but I don't think we can do it yet because we'd end up "applying the changes" twice. We'll need to make some modifications to freshMessageStore(...) and the callsites before we can get to that state. The goal of these first diffs is mainly to get assertMessageStoreMessagesAreEqual(...) "out of the way" to make room for future refactoring passes.