Brent Abrahams

Thanks for bringing this up. I see two issues: 1) If an error is thrown, everything in the transaction should be rolled back 2) Types aren't being passed through the...

Hi @tab00, I haven't forgotten about this. I've managed to reproduce the error using your code sample but haven't had time to roll out a fix. I can see the...

I've just released 0.7.5 on Atmosphere, which fixes the rollback issue. The type problem, however, isn't going to be fixed anytime soon. The object isn't a plain object -- it's...

There would need to be a hook at the right stage of the process to do this manually (i.e. using a function like the one above), and this hook doesn't...

No, I'm pretty sure a transform function wouldn't help in this case. The code needs to be refactored so that it records the type when the mutation data is "packaged...

@ericoe. So sorry I didn't reply here earlier: 1. The `description` you refer to would be the text that goes on the undo/redo buttons, right? `context` is a special-cased field,...

For 1., I agree that it's overkill. I'm thinking of introducing a `description` field for single-action transactions that auto-commit. Makes it quicker than wrapping the single mutator command in a...

Yes, you're right. That inverse is weird. I'll look into that. The fact that the transaction code just picks up the `_id` field shouldn't affect the update part -- as...

Haha... you're absolutely right. I wasn't thinking clearly about how the positional operator works. This is going to be difficult to support in the transactions package without a substantial refactor...

Yes, I'll certainly take a look at that weird inverse. I really need to write a bunch more tests for this package -- coverage is spotty at best ...