[dev] RDO describe() limitation

[Chuck Hagenbuch]

Quoting Duck <duck at obala.net>:

> A every day usage is to split data by tables. Manly in one2many way. I just a
> decision of what RDO is for. To low multiple column primary key or to force
> adding a resource wastefully special primary column just to identify your
> data. Even they can be in the other way. And believe me, no one will. As I
> see, no one will use RDO, even for its advantages, as he must reuse another
> DB access layer to update their data.
>
>> Not sure what you mean here. All of the relationship types work in RDO
>> right now, with single-keys for tables, at least in basic scenarios.
>
> Yes, when selecting only. But when you update a relationship if falls as it
> will need a primary key that not exist. (for example updating datatree node)

Can we talk about a concrete example? And not the datatree, because  
that's a bad design for scalability and the reason I'm moving towards  
RDO in the first place.

Also, btw, the datatree design doesn't allow updating any datatree  
attributes individually with any guarantees - it just deletes and  
re-writes them all. Not so great.

-chuck

-- 
"we are plastered to the windshield of the bus that is time." - Chris