couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Robert Samuel Newson <>
Subject Re: Descending replication
Date Thu, 04 May 2017 20:22:43 GMT

I agree this would be a great option, even a better default.


> On 4 May 2017, at 14:21, Johannes J. Schmidt <> wrote:
> Hi!
> I'm cross posting this issue from the PouchDB project:
> # Descending replication
> I'd like to discuss an idea that came up during Offline Camp: descending replication.
> Initial replication performance is still an issue, it's intrinsic to huge data sets.
> When you initially replicate its often the new data that's the most interesting to the
user. And that one arrives at latest. It would be good to be able to replicate in descending
order. That way the user would see new data instantly, while the rest drops in later.
> Also, the app could decide to stop replication at some point based on a memory limit.
Sometimes you don't want to sync an entire dataset (for example for event based data).
> Descending Replication would give us:
> 1. Relevant data first (improved UX)
> 2. Requests like "Give me 200MB of (latest) data"
> Do you see any generic problems? Wouldn't this be an easy change?
> Relax,
> Johannes

View raw message