On Thu, 15 Jun 2006 12:44:20 -0400 "Jorge Vargas" <[EMAIL PROTECTED]> wrote:
> On 6/15/06, Kevin Dangoor <[EMAIL PROTECTED]> wrote: > > > > > > One of the patches we have for SQLAlchemy 0.2 support in TG switches > > from using ActiveMapper for the identity classes to using the > > straight SQLAlchemy model. The argument that I can see for making > > that change is that ActiveMapper is an "extension", so it's not > > quite as supported as the standard data mapper approach. The > > disadvantage is that it's more verbose and may not be as easy to > > grok for new users. > > > > What opinions do you all have? > > > why is this based on identity only? I think we should scrap > ActiveMapper all together from TG , don;t get me wrong I like the > work Jonathan put in there but I think it hides the potential SA has > behind the fact of making it easy for new people. > > I believe the trade off between making it hard to start vs. making it > hard to make complex db stuff, is better to go complex at the start > because people could get flustracted saying, well TG is good for > small stuff but I can't do complex SQL on it. > > BTW, I'm also thinking that we should just remove the identity model > > classes from saprovider.py, since they're not working anyhow, and > > just have them in the model.py of quickstart. > > > yes they should be taken out, the only reason the ones in > soprovider.py are still there is for backward compatibility Maybe I'm mistaken, but I thought that if you used something like ActiveMapper, you could fall back to regular SA. Sort of the best of both worlds. If that's the case, in most cases, I prefer the ease at the beginning, but no hoops later on. Jason
signature.asc
Description: PGP signature
