On Wed, Mar 27, 2013 at 7:43 PM, Summers Pittman <supittma(a)redhat.com>wrote:
Sorry for the questionable formatting before, this should be more
So for offline and sync for 2.0 I’ve been doing some thinking/research
and come up with some broad topics to discuss so we can start honing in
on what we want it to do/look like.
This isn’t a spec, this isn’t a proposal, this is just trying to narrow
down what we want at a high level so we can pick things to focus on.
1. Documents vs Transactions
There are two “big picture” methods of doing sync. One is a Document
sync (Think like a gallery with videos and pictures). Documents are
saved, the whole document is sent to the server, then the whole document
is pushed out to other clients who are syncing against the same source.
The other is a Transactional sync where many small atomic operations are
sent to the server. The best analog I have is Google Drive/operation
Obviously the server side implementations of these are hilariously
divergent and I will leave the relative complexity of each as an
exercise for the list.
IMO, because Aerogear is a library we should focus on a Transactional sync
the users a more fine-grained control. But later, we should be able
to provide some wrappers around this atomic operations to offer something
close to document sync (or let the users easily define a "Document", like a
view in SQL)
2. Background vs Foreground sync
Does the application have to be opened (foreground) for syncing to
happen? As far as I know, native Web requires this (barring extensions
to the browser, plugins etc). I think Cordova can in Android, but I
haven’t researched it. iOS seems like a mixed bag, but generally you can
only sync if your application is in the foreground (but you can use
notifications and badges to communicate that there is a pending sync or
new data). I understand there is CoreData + iCloud that is supposed to
do something, but that seems like it is still foreground only. On
Android background sync is easy.
Is it OK to only have background syncing on some platforms but not others?
Shouldn't we trigger a sync only when we go from an offline status to an
online status ? And sync strategy should be configurable (both on client
side and server side)
3. Push vs Poll?
Obviously pushing updates is better for devices and users, but polling
will let things work better for legacy services which may not have push
support or which may be difficult to integrate into AG-Controller.
Should we support both on the clients?
Yes, we should promote push but have a fallback to a pull strategy
4. Multiple clients,multiple users, and conflicts
How do we want to support multiple users and multiple clients? How
should we try to do conflict resolution? What does authorization and
authentication look like here?
At a high level here are some options I have seen for conflict resolution:
A. Last in always wins. The server explicitly trusts things in the order
it gets and pushes that data out to users.
B. Clients are allowed one submit at a time and must wait for the server
to acknowledge the receipt. If there is a conflict the app can either a)
merge the data, b)reload the latest from the server and make the user do
his operation again, or c) create a new document and inform the user.
C. Operation Transforms. This was meant to solve the conflict and sync
problem. However it is a LOT of work
I would add another option :
E. The Strongest always win : i.e An admin user change will always be
considered with a higher weight when merging a conflict.
5. Offline Support
Really this is more what do we want to do for coming from an “offline”
mode to an “online” mode? Abstractly, operations which happen offline
are the same as operations which happen online just with a REALLY REALLY
laggy connection. :) We could just only viewing data when offline and
requiring a connection for editing, queueing an upload, etc.
Ideally, we should be able to provide the same functionality no matter we
are online or offline. For example, when I'm in the middle of the ocean, I
still want to be able to tag my Dolphins ;)
But we could think about "degraded service levels" ...
6. How much of this is the responsibility of AG-controller vs underlying
How should the controller expose resources to clients, how should the
controller send data to its underlying services, how much data should
the controller be responsible itself for?
Should it be easy for an Operation Transform system to integrate with
Should it be easy to write a Controller based project which polls a
third party source?
How would the server handle passing credentials to the third party source?
Appendix Use Cases:
Here are a few contrived use cases that we may want to keep in mind.
1. Legacy Bug Trackers From Hell
a. It is a webapp written in COBOL, no one will ever EVER update or
change the code
b. It has TONS of legacy but important data
c. It has TONS of users
d. It only has a few transactions per day, all creating and updating bug
e. Multiple users can edit the same report
2. Slacker Gallery
a. Each User has a multiple galleries, each gallery has multiple photos
b. A Gallery has only one user, but the user may be on multiple devices
c. Galleries may be renamed, created, and deleted
d. Photos may only be created or deleted. Photos also have meta data
which may be updated, but its creation and deletion is tied to the Photo
3. Dropbox clone
a. A folder of files may be shared among users
b. There is a size limit to files and how much storage may be used per
c. Files are not updated. If there is a new file, there is an atomic
delete and create operation
4. Email client
a. This is an AG-controller which accesses a mail account.
b. There are mobile offline and sync enabled clients which connect to
5. Google Docs clone
a. Operational Transform out the wazzoo
b. What would the server need?
c. What would the client need?
Appendix Reference (Open Source) Products:
Google Drive RealtimeAPI
Can you guys think of more projects/examples to look at for inspiration?
We could take a look at the Grails html5-scaffolding plugin which cover
some of this use cases
aerogear-dev mailing list