On Sun, May 27, 2012 at 10:15 PM, Paul DeBruicker
<[hidden email]> wrote:
Is there a strong reason to switch to 1.8 vs 1.7 other than the fact that 1.8 is the new hotness?
I remember that you made no guarantees but are Fuel 1.7 and Fuel 1.8 serialized formats compatible?
Indeed, they are not compatible. In fact, I guess all Fuel versions will be like that. It is really difficult not to change the format at all.
So...is it worth to update to 1.8? it depends on your needs. 1.8 provides new features (hooks) and some better performance.
- If you already have only a few data and it would be easy to migrate then I would do it.
- If it is too much data, then I would stay in 1.7 until you really need something from 1.8.
- If you need something of the new features/hooks of 1.8 or the speed is crucial, then I would update to 1.8.
If not is there a recommended process to switch from one format to the other?
I'd assume starting from a clean pharo you could:
1. load Fuel 1.7
2. materialize all the interesting Fuel 1.7 objects
3. removing Fuel 1.7
4. load Fuel 1.8
5. serialize all the interesting objects in Fuel 1.8
Exactly. Unfortunately, that's the only way we have to do it. Of course you can script that so it is not really a lot of "development time" but just the machine running ;)
I have not looked at the code but is there a way to do a streaming 'serialize to byte array' in conjunction with Zinc for populating a database?
Sorry, I couldn't follow here. All I can say is that you can pass to Fuel any stream (even Zinc ones). So you can do:
FLSerializer newDefault serialize: 'whatever' on: aZincWriteStream
you may need to implement some extension methods to Zinc streams. I think Nick already did them. If it is helpful, we can create a package Fuel-Zinc that adds such extension methods, just as we did for FuelFileSystem.
Thanks for your work on this.
--
Mariano
http://marianopeck.wordpress.com
Locked
