At 03:34 PM 10/9/2007 -0700, D John Anderson wrote:
Hmm. I think the repository is pretty separate from the app. For example, it would be pretty easy to replace the repository as long as the replacement had a similar API.
That's a bit like saying you could replace a bicycle with a motorcycle as long as you could find a motorcycle whose speed is controlled by pedalling. :)
Changing the API would be a problem if the replacement didn't provide the necessary features the app used.
Some repository features are used mainly by CPIA, for data that we don't necessarily need to *store* in the repository. If we stopped storing those things, we could use a replacement that traded off those features for better performance.
It's hard to imagine any data access mechanism whose API would not affect the code that used the API.
This isn't really the crucial issue; it's, "what parts of the code have to know about implementation details?" Currently, application code "knows" things it shouldn't have to know, regardless of the back-end, which makes experimenting with alternatives more difficult.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Open Source Applications Foundation "chandler-dev" mailing list http://lists.osafoundation.org/mailman/listinfo/chandler-dev
