Re: [Gnumed-devel] ID issuers, types, and validation

[Karsten Hilbert]

> What about a praxis
> 
> 1) modifying their GNUmed schema post-hoc, to add local constraints

Except foreign key constraints which would make the (1.4)
client refuse to run unless forced to.

> 2) keep track of such modifications, so as to be able to drop the extra local 
> constraints at the time of wanting to upgrade their GNUmed db
> 3) when wanting to upgrade their local db, they adjust their schema to match 
> what the GNUmed bootstrapper / updater requires
> 4) following the update, the praxis re-applies their local constraints
> 
> The above would not work the other way around. For example, if a praxis 
> *drops* base GNUmed constraints, and inputs data that GNUmed would normally 
> refuse, then they will be unable to re-apply the constraints because existing 
> data will be incompatible.
> 
> Doing it the other way around however -- making the data locally more 
> restrictive -- should not be a problem should it?

That's right. Off-hand I don't see a technical problem.

Karsten