Partial sync / TTL / pagination

[khalil-omer]

I'm wondering if there is a recommended pattern or planned feature to support syncing unbounded datasets.

For example, consider a Slack style app where I want to have a channels collection and a messages collection, but there's some complexity in the messages collection because I only want to sync the first page of messages for each channel. Until the user tries to scroll up in one particular channel. And then similarly, even if I only sync the first page of messages per channel, at some point I need to expire old messages that have not been rendered for some time, because the dataset will grow as new messages are sent/received.

I'm using Slack as an example, but this applies to many different use cases, especially those that are sort of 'time series' where navigating recent data is very common and should be very fast but where older data becomes less valuable fairly quickly.

[KyleAMathews]

👋 this is definitely a planned feature — we're hoping to have a design doc soon for people to look at.

[KyleAMathews]

@samwillis just put one up #315