Summary:
Until now, we've been hiding transactions beneath the surface. When you call persistModel, you're implicitly creating a transaction.
You could explicitly create them with `atomically`..., but there were several critical problems that are fixed in this diff:
- Calling persistModel / unpersistModel within a transaction could cause the DatabaseStore to trigger. This could result in other parts of the app making queries /during/
the transaction, potentially before the COMMIT occurred and saved the changes. The new, explicit inTransaction syntax holds all changes until after COMMIT and then triggers.
- Calling atomically and then calling persistModel inside that resulted in us having to check whether a transaction was present and was gross.
- Many parts of the code ran extensive logic inside a promise chained within `atomically`:
BAD:
```
DatabaseStore.atomically =>
DatabaseStore.persistModel(draft) =>
GoMakeANetworkRequestThatReturnsAPromise
```
OVERWHELMINGLY BETTER:
```
DatabaseStore.inTransaction (t) =>
t.persistModel(draft)
.then =>
GoMakeANetworkRequestThatReturnsAPromise
```
Having explicit transactions also puts us on equal footing with Sequelize and other ORMs. Note that you /have/ to call DatabaseStore.inTransaction (t) =>. There is no other way to access the methods that let you alter the database. :-)
Other changes:
- This diff removes Message.labels and the Message-Labels table. We weren't using Message-level labels anywhere, and the table could grow very large.
- This diff changes the page size during initial sync from 250 => 200 in an effort to make transactions a bit faster.
Test Plan: Run tests!
Reviewers: juan, evan
Reviewed By: juan, evan
Differential Revision: https://phab.nylas.com/D2353
Summary:
reduce scope of changes
more changes
Test Plan: Run 1 new test
Reviewers: juan
Reviewed By: juan
Differential Revision: https://phab.nylas.com/D2290
Electron 0.35.1 includes the tray fixes we contributed last week but also includes API restructuring and improvements. Most importantly, modules from electron are now imported via `require('electron')`
Summary:
This diff replaces the UnreadCountStore with a better approach that is able to track unread counts for all folders/labels without continuous (and cripplingly slow) SELECT COUNT(*) queries.
When models are written to the database, we currently don't send out notifications with the "previous" state of those objects in the database. This makes it hard to determine how to update counters. (In the future, we may need to do this for live queries). Unfortunately, getting the "previous" state is going to be very hard, because multiple windows write to the database and the "previous" state we have might be outdated. We'd almost have to run a "SELECT" right before every "REPLACE INTO".
I created an API that allows you to register observers around persistModel and unpersistModel. With this API, you can run queries before and after the database changes are made and pluck just the "before" state you're interested in.
The `ThreadCountsStore` uses this API to determine the impact of persisting a set of threads on the unread counts of different labels. Before the threads are saved, it says "how much do these thread IDs contribute to unread counts currently?". After the write is complete it looks at the models and computes the difference between the old count impact and the new count impact, and updates the counters.
I decided not to attach the unread count to the Label objects themselves because 1) they update frequently and 2) most things observing the DatabaseStore for categories do not care about counts, so they would be updating unnecessarily.
The AccountSidebar now listens to the ThreadCountsStore as well as the CategoryStore, and there's a new preference in the General tab for turning off the counts.
Test Plan: Tests are a work in progress, want to get feedback first!
Reviewers: juan, evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D2232
Summary:
We send database `trigger()` events through the ActionBrige to all windows of the app. This means that during initial sync, we're serializing, IPCing and unserializing thousands of models a minute x "N" windows.
This diff converts the payload of the trigger method into an actual class that implements a custom toJSON. It converts the impacted `objects` into a string, and doesn't deserialize them until it's asked.
Bottom line: this means that in many scenarios, we can avoid creating Contact models, etc. in composer windows only to broadcast them and then gc them.
Test Plan: No new tests yet, but this should definitel be tested. #willfix.
Reviewers: juan, evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D2236
Summary:
The Promise chain we were creating was never cleared and created a memory
leak. We instead use a `PromiseQueue` to cleanup finished promises.
Also added several more tests and verified that the memory leak is gone
with the Chrome profiler
Test Plan: new tests
Reviewers: bengotow
Reviewed By: bengotow
Differential Revision: https://phab.nylas.com/D2184
