On Tue, Jun 9, 2015 at 9:21 AM, Hilaire <hilaire@drgeo.eu> wrote:

Hi,

I want to use Fuel to persist a graph in file.

The classic use case is "modify some part of the model and persist immediately".

Now, it is not really efficient to re-serialize the whole graph each time, and I am wondering how you are approaching this particular point when using Fuel as your persistence back-end.

For example, various ideas coming to my mind: - Is it possible to update some part of an already serialized graph? - Should the graph be cut to pieces and serialized in several files? - Or is Fuel fast enough to just re-serialize the whole graph?

For Quuve app, when it runs on Pharo (for development and testing), we have a fuel backend and we do 2) and 3) of above. Quuve itself is split in different/isolated conceptual databases: clients, advisory, securities, system, application, etc. So each of them is a database. And yes, we have a fuel file per database and for each save we serialize the whole graph. You need to be careful about cross-references between databases.... Anyway, it works good enough for our scenario. Our database are around 5MB and the serialization is immediate.

How about you?

Thanks

Hilaire

-- Dr. Geo http://drgeo.eu http://google.com/+DrgeoEu

-- Mariano http://marianopeck.wordpress.com