Yeah there's a lot of interesting stuff being developed around pouchdb, I've been slowly checking it out the past couple of months:

Some beginning tutorials/intros to pouchdb: http://pouchdb.com/getting-started.html http://briantoth.github.io/Tutorial/2013/03/19/pouchdb-intro... http://mattgwwalker.wordpress.com/2013/04/08/pouchdb-2/

Todo Tutorial using AngularJS and PouchDb http://twilson63.github.io/ngTodoPouch/

Building Offline Applications With AngularJS and PouchDB http://mircozeiss.com/building-offline-applications-with-ang...

Other Apps Using PouchDB https://github.com/daleharvey/pouchdb/wiki/Who-or-What-Is-Us...

including this offline conference app https://github.com/axemclion/conference

PouchVision - inspect and interact w/pouch instances http://brudolph.pouchvision.jit.su/

Puton - bookmarklet for inspecting/editing pouch dbs http://puton.jit.su/

PeerPouch (PouchDB-over-WebRTC) - in development at one point https://github.com/natevw/pouchdb

Thats a great list, gonna add backbone-pouch (http://jo.github.io/backbone-pouch/) to it as well.

Thanks

I've been keeping an eye on PouchDB for a while; it looks like a lot of fun. How does authentication with the Couch server work, though? I don't see anything on the FAQ or after a brief perusal. CouchDB's apparently lax attitude toward security years ago was what kept me away, but I've always hoped/assumed that it improved since then. Has it?

CouchDB had some quirks related to user accounts as it has the added burden of trying to be an application server as well as a database server.

With PouchDB a lot of the application server logic doesnt apply and you can use CouchDB as a private data server that requires any connection to be authenticated, in the example we use basic auth but other methods are available and none of the security quirks with 'couchapps' apply.

And yes it is a lot of fun :)

> With PouchDB a lot of the application server logic doesnt apply

How do I keep one user from accessing another user's data?

By giving each user their own database :)

Sadly it's not possible to have CouchDB make that happen for you! You can very easily allow unauthenticated requests to create users, but there's no way to create those per-user databases without writing some trusted code that runs separately from CouchDB.

What will stop one user from accessing another user's database?

You can configure databases to require the correct user to authenticate before touching it..

I just put up this website, its ugly but working on the content, in particular I want it very easy to understand wtf it is and get started, would love to hear your feedback.

I especially like PouchDB-Server as a mini CouchDB replacement ( https://github.com/nick-thompson/pouchdb-server ). Congrats for implementing LevelDB into PouchDB so we can use it in Node.

The congratulations goes to https://twitter.com/chesles who implemented most of the levelDB adapter and Nick Thompson for the router + server implementation.

The server is really exciting, we mostly built it so we can reuse CouchDBs test suite however with it being so easy to install its becoming pretty invaluable

Here is daleharvey giving an intro to PouchDB at http://allyourbaseconf.com last year.

https://vimeo.com/56632201

Thanks for inviting me John, had a great time at the conference.

The website is mostly very clear but I wasn't sure what the db on the backend would be, it sounds like it expects to work with couchdb but an adaptor could potentially be made to work with anything server side?

Thats useful, I really should explain this a bit more on the site.

In the browser PouchDB will store its data in IndexedDB or WebSQL, in node its LevelDB, with the HTTP adapter you can use any product that implements the CouchDB HTTP Api protocol, currently this is CouchDB, Cloudant, PouchDB-Server and Couchbase Lite (previously TouchDB)

The important part is that it can only sync with a backend that implements the same replication protocol.

Dale - I've been really interested in this project since I work with CouchbaseLite. Hope you get good feedback. I'm particularly interested in setting this up with Sync_gateway. https://github.com/couchbaselabs/sync_gateway

PouchDB is part of the Sync Gateway test suite. So it should "just work"

I like the name.

hah thanks, I think Max Ogden came up with the name (this project existed a long time before I started working on it)

The example code is somewhat weird:

  var db = new PouchDB('dbname');
  
  db.put({
    _id: 'dave@gmail.com',
    name: 'David',
    age: 66
  });
  
  db.changes({
    onChange: function() {
      console.log('Ch-Ch-Changes');
    }
  });
  
  db.replicate.to('http://example.com/mydb');

Shouldn't you set up the event listener before "db.put()"?

I mostly just wanted to highlight the key features with 'real' code in the order you would come across them, you are right though might be worth switching that, thanks

You could define two functions to wrap the calls - do_a_put(), do_a_sync() - in that order and then call them in the reversed order in the example?

Is it possible to do something like this for a MySQL (innodb with relational keys) database ? I am about to begin work on a project that might be used in countries where we don't have constant connectivity (internet and electricity) .. and we don't want to lose or corrupt data due to connectivity ...

I mention this in the FAQ

Can PouchDB sync with MySQL / my current non CouchDB database?

No, the data model of your application has a lot of impact on its ability to sync, relational data with the existence of transactions make this harder. It may be possible given some tradeoffs but right now we are focussing on making PouchDB <-> (PouchDB / CouchDB) sync as reliable and easy to use as possible.

I've been using PouchDB for a month now, and it really is good. Simple, Yet powerful storage.

I'm not too familiar with Couch, but as I understand it caches views and updates them. Does Pouch do something similar, or does the query function filter all the documents in the database?

Currently PouchDB views do a full table scan, there is an open bug to do incremental views more inline with CouchDB - https://github.com/daleharvey/pouchdb/issues/99

I don't understand why they don't shim in using local storage for IE8 and 9?

Somewhere between IE support not being a hugely requested feature and that the limitations and speed of localStorage wouldnt make a very useful database. I think support will likely pop up at some point, it isnt very hard to write a new adapter, but right now I would rather focus on making it work very well in modern browsers.

I'm not sure how well localStorage would even work to be honest... as a key/value store though... prefixing "pouchdb_" + KEYNAME as the localstorage key would probably work. Notifications would probably be the biggest issue, I don't know pouch very well.

I just know that IE8 will probably be pretty relevant for another 2-3 years or so, and IE9 for a couple years after that. I don't like it any more than anyone else, but that's the world we live in.

Actually we do use localStorage for one thing, they support notifications :) IndexedDB (nor webSQL) have change notifications so we also do writes to a localStorage flag to make sure change notifications go through tabs.

a-la Cassandra, but under NodeJS?) I'm Jack's total lack of surprise!))

Its not really like cassandra, which has functionality aimed towards managing and querying huge volumes of data. The main feature here is the syncing, which is reasonably unique to the CouchDB world

just a simple question, why would you use this instead of, say, php+mysql ?

My main reason is for web applications that work fully offline and sync their data to all my devices.

Writing custom code to do that on a mysql stack is very hard.

Hi daleharvery quick question, if application is working fully offline, how is it syncing across all devices?

So you are on your iPad on a train. Your train goes into tunnel, you lose connection. The application still works because your data is saved in PouchDB.

You exit the tunnel, you gain connectivity PouchDB syncs your data to the CouchDB on the server. Other devices can now sync from the server.

It is able to work offline, but if a connection is available it syncs data (obviously the syncing doesnt work if there is no connection)

so what you're saying is, say you're working on your tablet/mobile/pc with an intermittent connection, it will continue working and update the db when a connection becomes available? is this the main benefit compared to the solution mentioned earlier?

Close but not quite, PouchDB is the database, you never need to be online to use it and it isnt queuing writes, its doing them locally.

The ability to sync is to let you use the same data across various devices.