Eating my own dogfood
We have been considering migrating Xaraya to the creole abstraction layer in the future. Creole is similar to the adodb library we use now. Originally the support for Creole was created to be able to implement a MS-SQL proof of concept quickly. After using the library for a bit I found it much easier to work with than adodb. Creole's API is much cleaner and it is PHP5 only, which gives us a head start for future Xaraya versions.
The Monotone setup we use allows us to test all this without disturbing main line development. Organizing these bigger changes into scenario's allows us to isolate these changes and see where they lead to. The decision to use or to throw away can thus be postponed to a time when we're ready to do so.
Since replacing the middleware is potentially a very big step, I noticed we (the Xaraya group) have some trouble making a decision on what to do with this. Therefore, the creole scenario has been merged into this site to show that a real life site (albeit a small one) can be migrated to creole without any effect. (Note: I've been running creole on other installs for a while now, so I've known this for a while)
Verdict
one change was necessary to wrap the 'GetRowAssoc' method present in adodb but not in creole. Everything else worked directly after the merge, so my preliminary conclusion if the decision hangs on 'what impact does this have?' is that the impact is very minimal. On top of that I think that the issues we might run into are not bad things as such, but chances to improve Xaraya.